Renewable Energy 30 (2005) 281–304

www.elsevier.com/locate/renene

Energy savings of office buildings by the use

of semi-transparent solar cells for windows
T. Miyazaki
, A. Akisawa, T. Kashiwagi
Graduate School of Bio-Applications and Systems Engineering, Tokyo University of Agriculture
and Technology, 2-24-16 Nakamachi, Koganei-shi, Tokyo 184-8588, Japan
Received 24 April 2004; accepted 29 May 2004

Abstract

The study investigated a PV window that consists of a double glazed window with semi-
transparent solar cells. The window provides natural light transmission as well as electricity
production. The effect of the PV window on energy consumption of office buildings was ana-
lyzed in terms of heating and cooling loads, daylighting, and electricity production. The pur-
poses of the study were to find the optimum solar cell transmittance and window to wall
ratio (WWR), and to estimate energy savings of the building. A standard floor of an office
building was modeled to run computer simulation, and annual energy simulation was per-
formed with EnergyPlus. The results showed that the solar cell transmittance of 40% and
WWR of 50% achieved the minimum electricity consumption in the building when artificial
lighting was controlled with daylighting. The optimum solar cell transmittance for PV
windows in different orientation was also presented. By using the optimum PV window, the
electricity consumption was reduced by 55% compared to the single glazed window with
WWR of 30% and no lighting control.
# 2004 Elsevier Ltd. All rights reserved.

Keywords: Energy simulation; Office buildings; Windows; Semi-transparent photovoltaics; Daylighting

1. Introduction

Energy consumption in commercial sector is growing in spite of the development

of energy efficient systems in Japan. This is mainly due to increasing floor area of
buildings year by year [1]. To reduce energy consumption in buildings, it is

Corresponding author. Fax: +81-42-388-7282.
E-mail address: [email protected] (T. Miyazaki).

0960-1481/$ - see front matter # 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.renene.2004.05.010
282 T. Miyazaki et al. / Renewable Energy 30 (2005) 281–304

necessary to contract energy demands as well as to install energy efficient supply

systems. In modern architecture, windows play an important role to reduce energy
demands in respect of heating and cooling loads and lighting requirement. There-
fore, the interrelation between window design and thermal performance of build-
ings has been extensively investigated [2–9].
By advancement of computational capability, the optimum window size and
types to minimize energy consumption of buildings have been explored by com-
puter simulation. The south window size, as well as building aspect ratio, to mini-
mize heating and cooling loads was analyzed by Mehlika et al. [2] From a
computer simulation of a residential building in five different climate regions in
Turkey, it was concluded that the south window size of 25%, which was required
in Turkish Insulation Regulation, was the optimum in hot climates. While, the
larger south window sizes were preferable in cold climates. Kontoleon and Bikas
[3] investigated the optimum glazed openings percentage from the viewpoint of the
indoor temperature and the energy consumption of the supply system. Al-Homoud
[4] carried out an optimization study on building design variables to minimize
annual energy consumption. Minimum glass area of 15% was the optimum in his
research except the cold climate where larger glass area was required to utilize solar
gain for heating. The economically optimum window size and orientation were
investigated by Johnson et al. [5]. The conclusion was that glass to wall ratio of
less than 20% resulted in the minimum life-cycle cost. In addition, the north orien-
tation was preferred for large glass to wall ratios. These precedent studies showed
that large window size resulted in the increase of a cooling load, while it causes the
decrease of a heating load because of enhanced solar gain.
The use of daylight is one of the most important factors to be taken into con-
sideration for window design. Daylighting affects heating and cooling loads of
buildings in terms of solar gain as well as heat gain from artificial lighting when
lighting control is installed. Bodart and De Herde [6] evaluated the impact of light-
ing energy savings on energy consumption of buildings by a combination of a day-
lighting simulation and a dynamic thermal simulation. The results showed that the
primary energy saving due to daylighting was around 40%. Sullivan et al. [7] inves-
tigated a method for an evaluation of fenestration and lighting system. The
relationship between solar aperture or effective daylighting aperture and electricity
use for cooling and for lighting was presented. A similar technique was used by
Lam and Li [8] for cooling dominated office buildings in Hong Kong. It was
shown that the peak cooling load and the annual electricity consumption were
reduced by 11% and by 13%, respectively, due to the use of daylighting. Zain-
Ahmed et al. [9] showed that the energy saving of 10% could be achieved by using
daylighting strategies in Malaysian buildings. Therefore, most researchers showed
that energy savings of 10–40% could be achieved by a daylighting scheme depend-
ing on the shape of buildings and climate zones.
Windows acquire an additional function of electricity production when they are
integrated with photovoltaics. The technology of semi-transparent photovoltaics
has been developed, and glass integrated photovoltaic modules have been applied
to glass facade or skylights [10,11], for example. To analyze the effect of photo-
 T. Miyazaki et al. / Renewable Energy 30 (2005) 281–304 283

voltaic glazing systems on energy consumption of buildings, a thermal aspect, an

optical aspect, and electricity production have to be examined comprehensively.
This type of approach has been practiced by Sylvester and Haberl [12], where an
energy simulation was performed for high-rise buildings in four cities in the US. A
photovoltaic glazed window with the transmittance of 40% was examined, and the
results showed that the PV window significantly decreased the electricity use of the
building. The optimum transmittance of the PV window was not investigated in his
research, however.
It is necessary to explore the optimum design for the evaluation of PV windows
because the thermal performance of the building and the electricity production
from the photovoltaics are interrelated. In the study, the application of windows
integrated with semi-transparent photovoltaics to buildings under the climate of
Japan was investigated by a computer simulation. The purposes of the study were
to find the optimum transmittance of the semi-transparent solar cell, and to esti-
mate possible energy savings of office buildings, by taking into consideration the
heating and cooling loads, daylighting, and electricity production.

2. Energy simulation of an office building

2.1. Heating and cooling loads calculation

Hourly heating and cooling loads were calculated by EnergyPlus [13], which is a
building energy simulation software developed by the US Department of Energy.
EnergyPlus calculates thermal loads of buildings by the heat balance method. The
heat balance method takes into account all heat balances on outdoor and indoor
surfaces and transient heat conduction through building fabric. It is more accurate
than the weighting factor method, which is used in precedent thermal loads calcu-
lation software such as DOE-2, because the heat balance method allows the vari-
ation of properties with time steps [14]. The simulation results of EnergyPlus have
been validated through analytical, comparative, and empirical tests [15,16].
Although EnergyPlus is capable of simulating heating, ventilation, and air con-
ditioning (HVAC) systems, the details of HVAC systems were not modeled since
the primary objective of the study was to examine the influence of PV windows on
thermal loads of buildings.
2.2. Description of the building simulated in the study

An office building in Tokyo, Japan was used to run the computer simulation.
The floor plan and the materials of the building elements were based on the Archi-
tectural Institute of Japan (AIJ) standard model [17] for thermal analysis of office
buildings. In the study, only the standard floor was modeled to reduce computa-
tional loads. The indoor temperatures of the upper and the lower floors were
assumed to be maintained at the same temperature as the standard floor. The floor
plan, the south side and east side elevations are depicted in Fig. 1. The floor has
four office zones, which are oriented to the northeast, northwest, southeast, and
284 T. Miyazaki et al. / Renewable Energy 30 (2005) 281–304

Fig. 1. Plan and elevation of the building.

southwest, respectively. The east side zones and the west side zones are separated
by a core zone. Each office zone is a square of 12 m in a side, and 3.6 m in height.
The dimension of the core zone is 9  24 m, and 3.6 m in height.
The properties of the wall, ceiling and floor materials are given in Table 1. The
values were obtained from Cymap’s software, QuickSlab. The estimated U-values
of the components were 0.80, 2.90, and 1.20 W/(mK) for the exterior wall, the
interior wall, and the ceiling/floor, respectively. In EnergyPlus, the heat transfer by
radiation, convection and conduction is calculated at each time step. The U-values
are not constant through the simulation because the radiative and convective heat
transfer is calculated by algorithms that take into account parameters such as tem-
perature difference between the surface and the air [18].
The office zones have windows on their exterior walls. Window to wall ratio
(WWR) of the office zones in the standard model is 30%. WWR was varied from
30% to 50% in parametric analysis.
Table 2 shows the design values of the internal heat gains, ventilation, and infil-
tration. The hourly variations of the internal heat gains were determined by the
simulation schedules [17] that are shown in Fig. 2. The ventilation works from 8:00
to 18:00 h, whereas the infiltration is uncontrollable.
 T. Miyazaki et al. / Renewable Energy 30 (2005) 281–304 285

Table 1
Walls, ceiling and floor construction (the order of the floor materials is the reverse of that of the ceiling
materials)
Layers Thickness Thermal cond. Density Specific heat
(outer to inner) (mm) (W/(mK)) (kg/m3) (J/(kgK))
Exterior wall
Cast concrete 150 1.4 2100 840
EPS 25 0.040 25 1400
Air space (Thermal resistance ¼ 0:18 m2  K=W)
Plasterboard 12 0.16 950 840
Interior wall
Cast concrete 150 1.4 2100 840
Ceiling/floor
Cast concrete 150 1.4 2100 840
Air space (Thermal resistance ¼ 0:18 m2  K=W)
Plasterboard 9 0.16 950 840
Fiberboard 12 0.06 300 1000

2.3. PV window

A typical PV window consists of two panes of glass with a semi-transparent pho-

tovoltaics layer between them. An amorphous silicon solar cell type PV window
was assumed in the study because of its aesthetic advantage for the application to
windows. Sanyo Electric Co., Ltd. has developed a semi-transparent amorphous
silicon solar cell, which is called the see-through a-Si solar cell [19]. The basic struc-
ture of the solar cell is the same as ordinary amorphous silicon solar cells. The
microscopic holes in the solar cell make the solar cell semi-transparent, however.
One of the advantages of the see-through a-Si solar cell is the natural light trans-
mission because the spectrum of the transmitted light is almost the same as that of
the incident light. The transmittance of the see-through a-Si solar cell is adjustable
by changing the area of the holes. Power output reduction of the solar cell, because
of the presence of the holes, is almost equal to the transmittance of the solar cell
[19]. One can find an application of the see-through a-Si solar cell to windows or
skylights in a few Japanese architecture [20,21].
The PV window investigated in the study consists of a double glazed window
with a see-through a-Si solar cell layer between the panes. The schematic of the PV

Table 2
Design values of internal heat gains, ventilation, and infiltration
Units Offices Core
2
Occupants (person/m ) 0.2 0.03
Lighting (W/m2) 25 15
Office equipment (W/m2) 20 –
Ventilation (m3/(m2h)) 4.0 0.6
Infiltration (air changes/h) 0.1 0.1
286 T. Miyazaki et al. / Renewable Energy 30 (2005) 281–304

Fig. 2. Hourly schedules of internal heat gains.

window is shown in Fig. 3. The transmittance of the solar cell was varied from
10% to 80% for parametric analysis.
To run the energy simulation by EnergyPlus, the optical properties of the glass
and the solar cell layer were required. Table 3 shows the optical properties of a
6 mm clear glass and a 10 mm clear glass that were obtained from the glass library
of WINDOW 5 [22]. For the solar cell layer, the following assumptions were made.

. The solar and visible transmittances are equal to the solar cell transmittance.
. The reflectance of the front side glass substrate is constant. In other words, the
sum of the transmitted light and the absorbed light is constant regardless of the
transmittance.
. The back side reflectance of the solar cell varies in response to the transmittance
of the solar cell. The reflectivity of the back side metal electrode is 90%.

Finally, the optical properties of the solar cell layer were determined as follows:
 T. Miyazaki et al. / Renewable Energy 30 (2005) 281–304 287

Fig. 3. Schematic of the PV window.

. The transmittance, sp ¼ 0:1 0:8.

. The front side reflectance, Rfp ¼ 0:1.
. The back side reflectance, Rbp ¼ ð1  sp Þ  0:9.

Electricity output from the solar cell was calculated from the electricity conver-
sion efficiency of the solar cell at standard test conditions, g, the glass layer trans-
mittance, sg, the solar cell absorptance, ap, the temperature coefficient of power
output, K, the effective solar cell area, Ap, the solar cell temperature, Tc, and the
solar radiation on the window, G. The electricity output, P, was calculated from
the following equation:

P ¼ Ggsg ap f1  KðTc  25ÞgAp ð1Þ

where the effective area was defined as Ap ¼ Aw ð1  sp Þ. Aw represents the window

area. The electricity conversion efficiency of 9% and the solar cell absorptance of
0.9 were assumed. The temperature coefficient of power output was assumed as

Table 3
Optical properties of glass
Glass type Solar Solar reflectance Visible Visible reflectance
transmittance transmittance
Front Back Front Back
Clear 6 mm 0.774 0.072 0.072 0.883 0.081 0.081
Clear 10 mm 0.698 0.066 0.066 0.861 0.080 0.080
288 T. Miyazaki et al. / Renewable Energy 30 (2005) 281–304

v
0.2%/ C [23]. The solar radiation on the window and the solar cell temperature
were obtained from the EnergyPlus output.
2.4. Daylighting and lighting control

The artificial lighting output to achieve a required illuminance level in a zone

can be decreased when daylight is available to a certain illuminance level. The con-
tinuous dimming control for artificial lighting was assumed in the study. The center
of each zone at desk height was used as a reference point of the interior illumi-
nance level, and the required illuminance level at the reference point was 700 lx
[24].
Any shading on windows reduces the available daylight. It is, however, usual to
use some shading device to avoid discomfort glare or to decrease excess solar heat
gain. An interior blind was modeled as a shading device. The blind was on when
the glare index at the reference point exceeded 22.
2.5. Electricity consumption of the heating and cooling systems of the building

It is necessary to estimate the electricity consumption of the heating and cooling

systems for the calculation of total electricity consumption of the building. The
electricity consumption of the system was calculated by simply dividing the heating
and cooling loads by COP of the system. An electric heat pump, which had the
heating and cooling COPs of 4.3 and 3.3 [25], respectively, was assumed as the
heating and cooling system of the building.

3. Results and discussion

3.1. Validation of the EnergyPlus results

Annual heating and cooling loads of the standard building model were calcu-
lated by EnergyPlus, and were compared with reference values [26] for the vali-
dation of the results. The window was a single glazed, and WWR was 30%. The
artificial lighting was not controlled with daylighting.
The monthly heating load and cooling load of the EnergyPlus results and the
reference values are shown in Figs. 4 and 5, respectively. In Japan, the heating load
occurs mainly from December to March, and the cooling load is substantial
between June and September. The simulation results conformed to this climatic
characteristics.
The heating and cooling loads of the core zone were notably smaller than the
office zones because of the less internal, fabric, and ventilation heat gains in
the summer, and the less fabric and ventilation heat losses in the winter. Although
the calculation results were less than the reference values at many calculated points,
the EnergyPlus results of office zones were reasonably close to the reference values,
and followed the trend of the monthly variation. The reason for the smaller values
of the calculation was that only the standard floor was modeled for the simulation.
Therefore, the heat losses and gains through roofs and ground floors were ignored.
T. Miyazaki et al. / Renewable Energy 30 (2005) 281–304 289

Fig. 4. Monthly heating load of the standard model.

Fig. 5. Monthly cooling load of the standard model.

290 T. Miyazaki et al. / Renewable Energy 30 (2005) 281–304

It was concluded that the simulation results from EnergyPlus were acceptable for
the purpose of the research, which was the comparison of the results between simu-
lation cases.
3.2. Parametric analyses on the solar cell transmittance and the window size

Parametric analyses on the solar cell transmittance and the window size were
carried out to investigate the effect of those parameters on the thermal perform-
ance of the building, and to find the optimum combination of the solar cell trans-
mittance and the window size. The solar cell transmittance was varied from 10% to
80%, and WWR was varied from 30% to 50%. The calculation was performed for
both the cases with and without lighting control. All the results including photo-
voltaic electricity output were presented in kilowatt-hours per unit floor area.

3.2.1. The effect on the heating and cooling loads

The hourly heating load and cooling load were obtained through annual simula-
tion, and their integrations were the annual total heating and cooling loads. Figs. 6
and 7 show the annual total heating load and the cooling load as a function of
solar cell transmittance, respectively. Without lighting control, higher solar cell
transmittance resulted in a smaller heating load and a larger cooling load because
of the increased solar gain through the window. With lighting control, the heating
and cooling loads did not follow this trend. With the increase of the solar cell
transmittance from 10% to a certain point, the heating load increased and the cool-

Fig. 6. The effects of the solar cell transmittance and WWR on the annual heating load.
 T. Miyazaki et al. / Renewable Energy 30 (2005) 281–304 291

Fig. 7. The effects of the solar cell transmittance and WWR on the annual cooling load.

ing load decreased because of the reduced heat gain from artificial lighting. The
trend was reversed, however, as the transmittance was further increased because
the influence of the solar gain became dominant.
The larger the WWR was, the larger the heating and cooling loads were. The
explanation is that the larger window area resulted in the larger U-value through
building fabric, which caused the increase of heat losses or gains. The larger win-
dow size also caused the augmentation of the solar heat gain in the summer.

3.2.2. The effect on the lighting electricity

Fig. 8 depicts the lighting electricity consumption as a function of solar cell
transmittance. Without lighting control, the lighting electricity consumption was
constant. When the artificial lighting was controlled with daylighting, the annual
total lighting electricity consumption was reduced as shown in Fig. 8. The lighting
electricity consumption decreased with the increase of the solar cell transmittance
and with the increase of WWR because of the enhanced daylight availability. The
amount of the reduction was, however, gradually diminished because there was no
potential of reduction once the required illuminance was achieved.

3.2.3. The effect on the electricity production from the photovoltaics

The effects of the solar cell transmittance and WWR on the electricity pro-
duction are shown in Fig. 9. The electricity production from the photovoltaics is
identical regardless of lighting control. The electricity production reduced linearly
when the solar cell transmittance was increased because the effective solar cell area
292 T. Miyazaki et al. / Renewable Energy 30 (2005) 281–304

Fig. 8. The effects of the solar cell transmittance and WWR on the annual lighting electricity consump-
tion.

Fig. 9. The effects of the solar cell transmittance and WWR on the annual electricity production from
the photovoltaics.
 T. Miyazaki et al. / Renewable Energy 30 (2005) 281–304 293

was reduced. The electricity production increased with WWR because of the exten-
sion of the effective solar cell area.

3.2.4. The effect on the total electricity consumption

A comprehensive outcome of the PV window was examined by the total elec-
tricity consumption, which was the summation of electricity consumption for the
electric heat pump and for lighting. To deduct the electricity production from the
photovoltaics, the electricity production was subtracted from the lighting electricity
consumption. Fig. 10 shows the annual total electricity consumption for the cases
with and without lighting control as a function of the solar cell transmittance. The
solar cell transmittance of 10% resulted in the minimum electricity consumption
when the artificial lighting was not controlled with daylighting. When the solar cell
transmittance was low, larger WWR resulted in smaller electricity consumption
despite the fact that larger WWR resulted in larger heating and cooling loads. This
trend was inverted at higher solar cell transmittances. The result implies that the
effect of the electricity production on the total electricity consumption was signifi-
cant with lower solar cell transmittances at which the electricity production was
large.

Fig. 10. The effects of the solar cell transmittance and WWR on the annual total electricity consump-
tion.
294 T. Miyazaki et al. / Renewable Energy 30 (2005) 281–304

With lighting control, the minimum electricity consumption was attained at dif-
ferent solar cell transmittances depending on WWR. The optimum solar cell trans-
mittances were 80% for WWR of 30%, 60% for WWR of 40%, and 40% for WWR
of 50%. The combination of the solar cell transmittance of 40% and WWR of 50%
achieved the minimum electricity consumption.
Fig. 11 shows a stacked bar chart indicating end-uses of electricity consumption
for WWR of 50% and the ‘‘with lighting control’’ case. The electricity production
from the photovoltaics was denoted as negative values. It could be observed that
the electricity consumption for lighting was the most significant factor among the
electricity end-uses. The variation of the electricity consumption for heating and
cooling was negligible through the transmittance of 10–80%. Therefore, the sum of
the heating, cooling and lighting decreased with the rise of the transmittance. The
electricity production reduced, however, at the same time, which resulted in the
appearance of the optimum point to minimize the total electricity consumption.
If other systems, such as gas boilers and absorption chillers, are used for heating
and cooling, the total primary energy consumption is different, and it can vary in
response to COP or efficiency of the system. The optimum solar cell transmittance
did not change, however, when the authors examined the systems of heat pumps
for cooling and boilers for heating, and that of absorption chillers and boilers.

Fig. 11. Electricity consumption by end-use as a function of the solar cell transmittance; WWR of 50%,
with lighting control.
 T. Miyazaki et al. / Renewable Energy 30 (2005) 281–304 295

3.3. Comparison with conventional window types

The PV window with the optimum solar cell transmittance was compared with a
single glazed and a double glazed windows in terms of the total electricity con-
sumption. The single glazed window consists of a 6 mm clear glass, and the double
glazed window consists of a 6 mm clear glass layer, 6 mm air space, and a 10 mm
clear glass layer from the outermost to the innermost.
The annual total electricity consumptions of three windows of the ‘‘without
lighting control’’ case were compared in Fig. 12. The details of each component are
shown in Table 4. The PV window achieved the smallest electricity consumption
among three window types because of the reduced cooling load and of the elec-
tricity production. Compared to the single glazed window, the double glazed win-
dow could reduce the total electricity consumption. The lower U-value of the
double glazed window caused the reduction of the heating load even though the
cooling load was slightly increased.
For the single and double glazed windows, the increase of WWR caused the rise of
the total electricity consumption. By contrast, the total electricity consumption of the
PV window fell with the increase of WWR because the electricity production was mag-
nified with WWR. The PV window with WWR of 50% could reduce the total electricity
consumption by 18% compared with the single glazed window with WWR of 30%.
The comparison between the single glazed, double glazed, and PV window under
the ‘‘with lighting control’’ case is shown in Fig. 13. The details are shown in

Fig. 12. The annual total electricity consumption with the electric heat pump system; without lighting
control.
296 T. Miyazaki et al. / Renewable Energy 30 (2005) 281–304

Table 4
Annual electricity consumption by window type; without lighting control
Units WWR 30% WWR 40% WWR 50%
SG DG PV SG DG PV SG DG PV
2
Heating (kW h/m ) 5.2 3.4 4.5 5.8 3.4 4.7 6.6 3.6 5.1
Cooling (kW h/m2) 12.1 13.0 10.2 12.9 14.2 10.6 13.3 15.2 10.9
Lighting (kW h/m2) 38.6 38.6 38.6 38.6 38.6 38.6 38.6 38.6 38.6
PV output (kW h/m2) 0 0 5.2 0 0 6.9 0 0 8.6

Table 5. It was remarkable that the lighting control considerably reduced the total
electricity consumption. The effect of WWR on the total electricity consumption
was similar to the ‘‘without lighting control’’ case. The PV window with WWR of
50% reduced the total electricity consumption by 13% compared to the single
glazed window with WWR of 30%. The reduction amounted to 54% when it was
compared to the single glazed with WWR of 30% and no lighting control.
3.4. Effect of zone orientation on the heating and cooling loads, lighting electricity,
and electricity production

The heating and cooling loads, the lighting electricity consumption, and the elec-
tricity production of each office zone were calculated, and the effect of zone orien-
tation was investigated for WWR of 50% with the lighting control.

Fig. 13. The annual total electricity consumption with the electric heat pump system; with lighting control.
 T. Miyazaki et al. / Renewable Energy 30 (2005) 281–304 297

Table 5
Annual electricity consumption by window type; with lighting control
Units WWR 30% WWR 40% WWR 50%
SG DG PV SG DG PV SG DG PV
2
Heating (kW h/m ) 7.1 5.0 5.1 7.6 4.9 5.3 8.5 5.1 5.8
Cooling (kW h/m2) 8.5 9.2 8.6 9.4 10.4 9.0 9.9 11.3 9.0
Lighting (kW h/m2) 13.9 14.5 15.6 13.3 13.6 15.5 13.1 13.4 16.6
PV output (kW h/m2) 0 0 1.2 0 0 3.1 0 0 5.8

3.4.1. The effect on the heating and cooling loads

Figs. 14 and 15 show the heating load and the cooling load as a function of solar
cell transmittance, respectively. The difference between the northern zones and the
southern zones was large, and was as much as 32% for the heating load at the solar
cell transmittance of 80%. The difference between the eastern zones and the western
zones was small, which indicated that the direct solar gain in the morning was
almost equivalent to that in the evening. The maximum heating load occurred at
the transmittance of 20% for the southern zones, while it occurred at the transmit-
tance of 40% or 50% for the northern zones. The minimum cooling loads also
occurred at different solar cell transmittances, and were 20% or 30% for the south-
ern zones, and 30% or 40% for the northern zones.
The different transmittances at the maximum or minimum points were arisen
from the significance of the direct solar gain through the window. In the southern

Fig. 14. The annual heating load for each office zone; WWR 50%, with lighting control.
298 T. Miyazaki et al. / Renewable Energy 30 (2005) 281–304

Fig. 15. The annual cooling load for each office zone; WWR 50%, with lighting control.

zones, the direct solar gain overrode the reduction of heat gain from lighting at the
solar transmittances of more than 30%. In the northern zones, the influence of the
direct solar gain turned dominant at higher transmittance since the northern zones
received less direct solar radiation.
3.4.2. The effect on the lighting electricity consumption
The lighting electricity consumption is presented in Fig. 16. When the solar cell
transmittance was less than 40%, the difference between the northern zones and the
southern zones was more than 15%. The difference was attenuated with the
increase in transmittance, and it was negligible at the solar cell transmittance of
80%.
3.4.3. The effect on the electricity production from the photovoltaics
Fig. 17 depicts the electricity production from the photovoltaics as a function of
solar cell transmittance. As for the electricity production from the photovoltaics,
the difference between the eastern zones and the western zones was negligible, while
the difference between the southern zones and the northern zones was significant.
This is obviously due to the amount of direct solar radiation. The difference
between the southern zones and the northern zones was about 33% at any trans-
mittances.
3.4.4. The effect on the total electricity consumption
Fig. 18 shows the total electricity consumption as a function of solar cell trans-
mittance, where the electric driven heat pump was assumed as the heating and
 T. Miyazaki et al. / Renewable Energy 30 (2005) 281–304 299

Fig. 16. The annual lighting electricity consumption for each office zone; WWR 50%, with lighting con-
trol.

cooling system. The electricity consumption of the southern zones was smaller than
the northern zones by more than 20% when the solar cell transmittance was
between 10% and 40%. The minimum electricity consumption was achieved at the
transmittance of 40% for the SW Zone and the SE Zone, at the transmittance of
50% for the NW Zone, and at the transmittance of 60% for the NE Zone.
The results implied that further reduction of the total electricity consumption
would be possible by the use of appropriate solar cell transmittances for PV win-
dows corresponding to the orientation. The optimum solar cell transmittances for
PV windows of different orientation in each zone were investigated by parametric
analysis, and are described in the next section.

3.4.5. The optimum solar cell transmittance of the PV window of different orientation
Each zone except the core zone has PV windows of two different orientations on
the exterior walls. Parametric analysis on the solar cell transmittance of the PV
windows in each zone was carried out. The transmittance was varied from 10% to
80%, and WWR was constant at 50%. The lighting was controlled with day-
lighting.
Fig. 19 depicts a three-dimensional graph of the total electricity consumption as
a function of the solar cell transmittances of the south and east side windows in the
SE Zone. Contours are shown on the surface in the graph. It was shown that the
total electricity consumption was less than 22 kW h/m2 when the transmittance of
the south side window was 30–40% and the transmittance of the east side window
300 T. Miyazaki et al. / Renewable Energy 30 (2005) 281–304

Fig. 17. The annual electricity production from the photovoltaics for each office zone; WWR 50%, with
lighting control.

was around 50%. The minimum electricity consumption in the SE Zone was
achieved with the south side transmittance of 30% and the east side transmittance
of 50%.
The optimum transmittances, which minimize the total electric consumption, in
the other zones could be obtained by similar parametric analyses. The optimum
solar cell transmittance in each office zone was summarized in Table 6. The results
revealed that the optimum transmittances of the southern zones were the same.
Those of the northern zones were also the same as each other.
The optimum solar cell transmittance of the south side PV window was 30%.
The optimum transmittance of the south side PV window was lower than that of
the other side PV windows because the south side window received the largest
annual solar radiation to produce electricity. It also had an effect of the reduction
of the excess solar heat gain. The optimum transmittance of the north side PV win-
dow was 80%. The transmission of daylight was more beneficial to the north side
window because the north side window receives not direct solar radiation but dif-
fuse solar radiation. It was also notable that the optimum transmittance of the east
and west side PV windows in the southern zones was different from that in the
northern zone. The transmittance of the east and west PV windows in the northern
zones was low because the north side window had high transmittance to provide
 T. Miyazaki et al. / Renewable Energy 30 (2005) 281–304 301

Fig. 18. The annual total electricity consumption with the electric heat pump system for each office zone;
WWR 50%, with lighting control.

daylighting. In contrast, the southern zones accepted more daylighting through the
east and west PV windows to compensate small daylight transmission through the
south side PV window.

Fig. 19. The effect of the solar cell transmittance on the total electricity consumption of the SE Zone.
302 T. Miyazaki et al. / Renewable Energy 30 (2005) 281–304

Table 6
The optimum solar cell transmittance of the PV window in each zone
South/North side (%) East/West side (%)
SE Zone 30 50
SW Zone 30 50
NE Zone 80 30
NW Zone 80 30

By the design with the optimum solar cell transmittance in each zone, the total
electricity consumption of the building was reduced by 2.4% more compared to the
best case of the uniform transmittance design, which was WWR of 50% and the
transmittance of 40%. The reduction amounted to 55% compared to the standard
model.

4. Conclusions

In the study, the application of the see-through solar cell to windows of office
buildings was investigated. The parametric analyses on the solar cell transmittance
and on WWR were carried out, and the optimum values to minimize the annual
total electricity consumption were found. The energy saving by the use of PV win-
dow was also estimated taking into account the effect of daylighting. The remark-
able findings are listed below.

. Without the lighting control, smaller cell transmittance gives less electricity con-
sumption irrespective of WWR.
. With the lighting control, the optimum solar cell transmittances were 80% for
WWR of 30%, 60% for WWR of 40%, and 40% for WWR of 50%.
. The combination of the solar cell transmittance of 40% and WWR of 50%
achieved the minimum primary energy consumption in the case of uniform
transmittance for all window orientation. The energy saving of 54% was
achieved compared to the standard model.
. The total electricity consumption was significantly reduced by the lighting control.
. A 2.4% more reduction was attained by the optimum design of the solar cell
transmittance of each zone compared to the design with uniform transmittance
at 40% and WWR of 50%. The reduction was 55% compared to the standard
model.

It should be noted that the study did not consider the influence of shading by
surrounding buildings, which would reduce the benefit of daylighting and the pho-
tovoltaic output. The results showed, however, possible energy savings under ideal
conditions.
The total electricity consumption calculated in the study assumed a constant
COP. Our work should incorporate HVAC systems for the analysis of energy
efficient supply systems.
 T. Miyazaki et al. / Renewable Energy 30 (2005) 281–304 303

References
[1] Guidance of the energy saving standard of buildings and calculation. Institute for Building
Environment and Energy Conservation, 2000.
[2] Mehlika N, Inanici F, Nur D. Thermal performance optimization of building aspect ratio and
south window size in five cities having different climatic characteristics of Turkey. Building and
Environment 2000;35(1):41–52.
[3] Kontoleon KJ, Bikas DK. Modeling the influence of glazed openings percentage and type of glaz-
ing on the thermal zone behavior. Energy and Buildings 2002;34(4):389–99.
[4] Al-Homoud MS. Optimum thermal design of office buildings. International Journal of Energy
Research 1997;21(10):941–57.
[5] Johnson CA, Besant RW, Schoenau GJ. Economic preferred window orientation and optimum fen-
estration design of a non-daylit and daylit large office building for different climate conditions and
different billing structures. ASHRAE Transactions 1990;96(1):23–33.
[6] Bodart M, De Herde A. Global energy savings in offices buildings by the use of daylighting. Energy
and Buildings 2002;34(5):421–9.
[7] Sullivan R, Lee ES, Selkowitz S. A method of optimizing solar control and daylighting perform-
ance in commercial office buildings. Proceedings of ASHRAE/DOE/BTECC Conference on the
Thermal Performance of the Exterior Envelopes of Buildings V. 1992, p. 313–9.
[8] Lam JC, Li DHW. An analysis of daylighting and solar heat for cooling-dominated office build-
ings. Solar Energy 1999;65(4):251–62.
[9] Zain-Ahmed A, Sopian K, Othman MYH, Sayigh AAM, Surendran PN. Daylighting as a passive
solar design strategy in tropical buildings: a case study of Malaysia. Energy Conversion and Mana-
gement 2002;43(13):1725–36.
[10] Chehab O. The intelligent façade photovoltaic and architecture. Renewable Energy 1994;5(1):188–
204.
[11] Sick F, Erge T, editors. Photovoltaics in buildings. International Energy Agency; 1996.
[12] Sylvester KE, Haberl JS. The effects of PV glazing on the energy consumption of high rise commer-
cial buildings. Proceedings of Solar Engineering 2000, A part of Solar 2000: Solar Powers Life,
Share the Energy. 2000, p. 217–27.
[13] Crawley DB, Lawrie LK, Winkelmann FC, Buhl WF, Huang YJ, Pedersen CO, et al. EnergyPlus:
creating a new-generation building energy simulation program. Energy and Buildings 2001;33(4):
319–331.
[14] Strand R, Winkelmann F, Buhl F, Huang J, Liesen R, Pedersen C, et al. Enhancing and extending
the capabilities of the building heat balance simulation technique for use in EnergyPlus. Proceed-
ings of Buildings Simulation’99, IBPSA, vol. II. 1999, p. 217–27.
[15] Witte MJ, Henninger RH, Glazer J, Crawley DB. Testing and validation of a new building energy
simulation program. Proceedings of Building Simulation. IBPSA; 2001.
[16] Olsen EL, Chen QY. Energy consumption and comfort analysis for different low-energy cooling
systems in a mild climate. Energy and Buildings 2003;35(6):561–71.
[17] Takizawa H. Proposal of the standard problem for office buildings. The 15th Heat Symposium of
Architectural Institute of Japan. 1985, p. 35–42.
[18] EnergyPlus Engineering Document. The US Department of Energy, 2003.
[19] Takeoka A, Kouzuma S, Tanaka H, Inoue H, Murata K, Morizane M, et al. Development and
application of see-through a-Si solar cells. Solar Energy Materials and Solar Cells 1993;29(3):243–52.
[20] Fukai K. Solar power generation system in Ota city hall, development of a seethrough-type
amorphous module in laminated glasses. Journal of the Institute of Electrical Installation Engineers
of Japan 2000;20(6):427–9.
[21] Sato A. Semitransparent solar battery system at Tamatukuri-onsen ‘‘yuyu’’. Journal of the Illumi-
nating Engineering Institute of Japan 1998;82(9):756–8.
[22] WINDOW 5.0. User manual. Lawrence Berkeley National Laboratory, 2001.
[23] Oshiro T, Nakamura H, Imataki M, Sakuta K, Kurokawa K. Practical values of various para-
meters for PV system design. Solar Energy Materials and Solar Cells 1997;47(1–4):177–87.
304 T. Miyazaki et al. / Renewable Energy 30 (2005) 281–304

[24] Kiya H. et al. (eds) Heating, air-conditioning and sanitary engineering handbook. 12th ed., vol. 3.
The Society of Heating, Air-Conditioning and Sanitary Engineers of Japan, 1995. p. 49.
[25] Nagano H. et al. (eds) JAR Handbook. 5th ed., vol. 3. Japanese Association of Refrigeration, 1993.
p. 83.
[26] Ishifuku A. et al. (eds) City gas driven cogeneration system plan, design, and evaluation. The
Society of Heating, Air-Conditioning and Sanitary Engineers of Japan, 1994 [Chapter 4].