Solar Energy 80 (2006) 65–77

www.elsevier.com/locate/solener

Photovoltaic collectors efficiency according

to their integration in buildings
Alain Guiavarch, Bruno Peuportier *

Centre Energétique et Procédés, Ecole des Mines de Paris, 60, Bd St. Michel, 75272 Paris Cedex 06, France

Received 13 December 2004; received in revised form 24 June 2005; accepted 5 July 2005
Available online 22 August 2005

Communicated by: Associate Editor Arturo Morales-Acevedo

Abstract

A model for building integrated photovoltaic systems has been developed and implemented in a dynamic simulation
tool. This tool takes into account the thermal interactions between the PV collector and the building. The influence of
the type of integration upon the PV collector efficiency has been evaluated and hybrid PV/air collectors have been stud-
ied. An overall efficiency is defined, including the production of electricity and heat. A case study has been performed on
two different typical buildings. In the case of a multi-crystalline silicon PV collector integrated on the roof of a single
family house located in Paris, the efficiency of unventilated PV modules fixed on the roof is 14%. If the PV collector is
used to preheat the ventilation air, the efficiency reaches 20%. A proper building integration also improves the environ-
mental balance of PV technologies over their life cycle.

2005 Elsevier Ltd. All rights reserved.

Keywords: Building integrated photovoltaic; Simulation tool; Environmental life cycle assessment

1. Introduction point of view, this can be achieved by improving the effi-

During the last years, the integration of photovoltaic
(PV) collectors in buildings has progressed and new
components have been successfully implemented in
architecture (IEA, 1996; IEA, 2000). PV modules can
be used as a cover material onto a roof or a façade, inte-
grated in chimneys or over a roof. Semi-transparent col-
lectors can be used to cover an atrium. Beyond the
aesthetic aspect, the aim of building integration is to re-
duce the pay back time of PV systems. From the energy
*
Corresponding author. Tel.: +33 1 4051 9151; fax: +33 1
4634 2491.
E-mail address: bruno.peuportier@ensmp.fr (B. Peuportier).
ciency of PV collectors.
The aim of this paper is to show how the type of inte-
gration can influence the efficiency of PV collectors.
A brief description of the simulation tool will be gi-
ven first. Then we focus on the modelling of building
integrated PV collectors. At last, the application of the
model will be illustrated by studying PV collectors inte-
grated in two typical residential buildings in France.
2. The simulation tool
The thermal simulation tool of multi-zone buildings
named COMFIE (Peuportier and Blanc Sommereux,

0038-092X/$ - see front matter
2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.solener.2005.07.004
66 A. Guiavarch, B. Peuportier / Solar Energy 80 (2006) 65–77

Nomenclature

Cf primary energy conversion factor, – U heat loss coefficient, W m
2 K
1

Cp specific heat, J kg
1 K
1 V voltage, V
E energy, kW h w width, m
G global irradiance, W m
2 a absorption factor, –
H irradiation, kW h g efficiency, –
h heat transfer coefficient, W m
2 K
1 n transparency, –
I current, A s transmittance, –
k Boltzmann constant, J K
1
L height, m Subscripts
m_ mass flow rate, kg s
1 ext external
Noct normal operating cell temperature, K f fluid
Q_ heat flow rate, W ov overall
q00 heat flux, W m
2 pv photovoltaic
q elementary charge, eV th thermal
R thermal resistance, W
1 m2 K z, adj adjacent zone
T temperature, K

1990) allows heating and cooling loads as well as tem-
perature profiles in different zones to be evaluated. The
program has been developed using an object oriented
approach, allowing modules to be linked to the core of
the program. These modules can represent solar compo-
nents, and especially building integrated photovoltaic
systems.
During the simulation, parameters are exchanged be-
tween each module and the core of the program: zone
temperatures and energy produced by the PV collector
for example (see Fig. 1). In this way, the thermal interac-
tion between photovoltaic collectors and the building can
be taken into account. The heat capacitance of the PV
collector is considered negligible compared to the thermal
mass of the building walls. Hourly climatic data is used,
but a smaller time step (e.g., 15 min) is generally needed
to model the dynamic behaviour of the building, the out-
put being integrated over 1 h. Faster dynamic evolution
(e.g., sun and cloud alternation) is not accounted for.
The simulation tool can be described according to its
three essential steps: input, dynamic simulation, and
output.
Module 1
First, the building envelope model is defined by its
geometry and elements (wall compositions, window
types, etc.). If a photovoltaic collector is integrated into
the envelope, its thermo-physical parameters must be
taken into account. The main parameters are the
emission and absorption coefficients, and for a PV glaz-
ing the U (heat losses) and g (solar transmittance)
values.
After describing the building envelope, the photovol-
taic system is defined. The system consists of two com-
ponents: the PV collector itself, and the inverter.
Although the simulation tool is able to simulate stand-
alone PV systems, we focus in this article on grid-con-
nected systems (i.e., without battery). As developed in
the next paragraph, electrical, geometrical and thermo-
physical parameters are needed.
Once the project is defined, the simulation is per-
formed using hourly meteorological data. Concerning
the building, the program calculates the energy loads
of the building, and the hourly temperatures in the dif-
ferent zones. Concerning the PV system, the program
calculates the energy produced by the system, and can
generate hourly outputs like the electricity produced
and the incident radiation on the PV collector.

Building
Simulation :
zone temperatures 3. Building integrated photovoltaic systems modelling
COMFIE
Module n :
heat production PV systems The model presented in this paper has been devel-
oped with the aim of taking into account many types
of integration, according to three main classes of mod-
space heating demand els: integration without thermal interaction with the
electricity production
building, integration without air gap and integration
Fig. 1. Overview of the simulation tool. with ventilated air gap.
 A. Guiavarch, B. Peuportier / Solar Energy 80 (2006) 65–77
3.1. Integration without thermal interaction
This type of model corresponds to PV collectors
which are not integrated into the building envelope,
for instance on a terrace roof (Fig. 2), or used as
shading.
The electricity produced by the PV collector is calcu-
lated using the Ôone-diode modelÕ (Ricaud, 1997). The
current at the load I is given by the characteristic equa-
tion as follows:
   
q V þ IRs
I ¼ I L
I 0 exp ðV þ IRs Þ
1
; ð1Þ
ckT J Rsh
where Rs, Rsh and c are parameters characterizing the
PV collector. The photocurrent IL is assumed to be pro-
portional to the incident radiation. According to Eq. (1),
the available power, which is given by multiplying the
current I by the voltage V, is function of the junction
temperature TJ. This parameter is the operating temper-
ature of the PV cells, and is assumed to be uniform in the
whole PV collector. This temperature is calculated
according to Duffie and Beckman (1991). The energy
balance of a unit area of module which is cooled by
losses to the surroundings can be written as
ðspv apv ÞG ¼ gpv G þ U pv ðT J
T ext Þ;
where spv is the transmittance of the cell cover, apv is the
absorption factor on the surface of the cells, gpv is the
electrical efficiency of the PV collector, Text is the exter-
nal temperature, G is the global irradiance incident on
the module and Upv is the heat loss coefficient on both
sides of the PV module (including radiative and convec-
tive transfers).
If we introduce the nominal operating cell tempera-
ture (Noct), defined as the cell or module temperature
that is reached with a solar radiation level GNoct of
800 W/m2, a wind speed of 1 m/s, an ambient tempera-
ture TaNoct of 20 C, and no load operation (i.e.,
gpv = 0), the junction temperature can be expressed as
follows:
   outside) is linearised and integrated in the global surface
Noct
T aNoct gpv
T J ¼ T ext þ G 1
. ð3Þ
GNoct ðspv apv Þ
PV collector
Building wall
Fig. 2. Example of integration without thermal interaction.
67
In the former equation, TJ is a linear function of the
external temperature Text and of the incident irradiance
G.
3.2. Integration without air gap
PV modules are sometimes placed in front of a wall
or a roof. In such a case, the PV modules are not cooled
like in the previous case and this influences their effi-
ciency (depending on the technology).
The PV collector is modelled as a material layer in
the composition of the wall on which it is fixed. This im-
plies to determine the different characteristics of the PV
collector listed hereunder:
• density qpv;
• specific heat C ppv ;
• thermal conductivity kpv;
• thickness epv;
• absorption coefficient apv;
• emittance epv.
But in most cases, the thermal resistance, and espe-
cially the heat capacitance (Jones and Underwood,
2001) of the PV collector can be neglected compared
ð2Þ to the other materials constituting the wall.
In order to calculate the junction temperature of the
PV collector TJ, it has been shown that a one dimension
steady state model performs almost as good as more de-
tailed models (Zondag et al., 2002). It means that it is
possible to calculate TJ using a simple thermal balance,
as described in Fig. 3.
The expression for TJ is similar to (3). The thermal
resistances R1 and R2 are calculated according to the
type and the characteristics of the PV module and the
building wall. The surface heat transfer coefficients be-
tween the PV collector and the building wall, and both
adjacent zones (hext and hz,adj) depend on the slope of
the PV element, the wind exposure (convective transfer)
and the emittance of the surface (radiative transfer). The
radiative transfer (e.g., longwave radiation loss to the
heat transfer coefficients hext and hz,adj.
We can notice that the energy balance is made
according to a steady state model. Nevertheless, the
thermal mass of the wall is taken into account via the
adjacent zone temperature Tz,adj. This temperature is
Ext. side PV Int. side Building
Outside zone
surface cells surface
(τ PVαPV-ηPV)G
Text TJ Tz,adj
1/hext R1 R2 1/hz,adj
Fig. 3. Energy balance for the PV wall.
68 A. Guiavarch, B. Peuportier / Solar Energy 80 (2006) 65–77

calculated at each time step by the core of the program with

(COMFIE), and the thermal mass is accounted for: a fi- hC1
nite volume model is created, and reduced using modal h01 ¼ ;
1 þ R2 hC1 þ hC1 1þRhext1 hext
analysis.  
1 þ R1 hext 1 þ R1 hext 00
T 01 ¼ T ext þ S 1
R2 þ qr ;
3.3. Integration with ventilated air gap hext hext
ð5Þ
hC2 hz;adj
h02 ¼ ;
This class of model applies if the PV collector is hC2 þ hz;adj
cooled by air circulation: heat can be recovered by form- 1 1 00
ing an air collector. The model developed applies for air T 02 ¼ T z;adj þ S2 þ q
hz;adj hz;adj r
collectors in general (Fig. 4). The PV collector may con-
stitute the cover of the air collector, or in some cases the and
absorber (leading to a higher cell temperature). 
S 1 ¼ ð1
nÞðspv apv
gpv Þ þ nag G;
The thermal resistance network shown in Fig. 5 ap-
plies if the PV collector acts as a cover (a similar thermal S 2 ¼ nsg aa G; ð6Þ
resistance network is used if the PV collector acts as an q00r ¼ hrad ðT 1
T 2 Þ;
absorber). The parameter n is the transparency of the PV
where sg and ag are the transmittance and the absorption
collector: for an opaque PV collector n = 0, and for a
factor of the glazing without PV cells, and aa the absorp-
semi-transparent n is the ratio of the area not covered
tion factor of the absorber.
by the PV cells. Heat is transferred to the flowing air
Applying energy conservation to a differential control
(at a temperature Tf) by convection from both the PV
volume (see Fig. 6), we obtain:
cover (via the convective exchange coefficient hC1) and
the absorber surface (via hC2). The radiative heat trans- dT f 


_ p
mC ¼ w h01 T 01
T f þ h02 T 02
T f ; ð7Þ
fer between the absorber and the cover is modelled using dx
a coefficient hrad (linear approximation). The tempera- where m_ is the air mass flow rate, Cp the specific heat of
ture Tf is assumed to vary only in the flow direction. the fluid, and w the width of the air collector.
According to the scheme presented in Fig. 5, the heat The previous equation can be integrated:
flux q00 produced by the collector and transferred to the  

building via the air flow is given by w h01 þ h02
T f ðxÞ ¼ T f;in þ 1
exp
x


mC
_ p
q00 ¼ h01 T 01
T f þ h02 T 02
T f ; ð4Þ  0 0


h1 T 1
T f;in þ h02 T 02
T f;in
 ; ð8Þ
h01 þ h02
where Tf,in is the inlet temperature of the fluid.
The heat flow rate Q_ transferred to the fluid is given
by
Q_ ¼ mC
_ p ðT f;ou
T f;in Þ; ð9Þ
where Tf,ou, is the outlet temperature of the fluid, given
by Eq. (8). Finally, replacing (Tf,ou
Tf,in) in Eq. (9) by
the expression from Eq. (8), we obtain:
_ p
mC  
Fig. 4. Air collector.
Q_ ¼ ð1
e
k Þ S 0
hL ðT f;in
T 0ext Þ ; ð10Þ
hL

Fig. 5. PV/air collector thermal resistance network; the thermal resistance of the insulation of the building wall is included in the
coefficient hz,adj.
 A. Guiavarch, B. Peuportier / Solar Energy 80 (2006) 65–77 69

Tf + dTf

dx x

Tf

Fig. 6. Control volume in the air gap.

Fig. 7. Semi-transparent PV cover with PV cells on the external

where
pane (a) or on the internal pane (b) of the double glazing.
hL ¼ h01 þ h02 ;


h0 þ h02 wL
k¼ 1 ; ancy forces are balanced by the pressure drops due to
_ p
mC
h0 ð1 þ R1 hext Þ h0 friction yields a third order polynomial equation (see
S0 ¼ 1 S1 þ 2 S2; Eq. (13)). The polynomial equation shown below is valid
hext hz;adj
0
for laminar flow, and a similar equation is found for tur-
h h0 bulent flow.
T 0ext ¼ 0 1 0 T ext þ 0 2 0 T z;adj
h1 þ h2 h þ h2
 01    Am_ 3 þ Bm_ 2 þ C m_ þ D ¼ 0. ð13Þ
1 h2 1 þ R1 hext 0 00
þ 0
R2 þ h1 qr The entrance and exit loss coefficients, and the friction
h1 þ h02 hz;adj hext
phenomenon are taken into account in coefficients A
ð11Þ
and B. The heat transferred to the air flow, the air prop-
and L is the height of the collector. erties (density and dilatation factor), and different geo-
The thermal efficiency of the PV/air collector can be metrical parameters of the collector (tilt angle, cross-
derived: section, length and hydraulic diameter) are accounted
for in coefficient D. The wind velocity and pressure effect
Q_
gth ¼ . ð12Þ are taken into account in C. The model is valid for nat-
ðwLÞG
ural and mixed convection.
The convective heat transfer coefficients hC1 and hC2 The heat flow rate transferred to the air flow in the
are calculated using correlations given by Kakac et al. collector Q_ can be deduced from the heat balance de-
(1987) for a laminar flow and by Kays and Leung tailed above. The convective heat transfer coefficients
(1963) for a turbulent flow and for forced convection. hC1 and hC2 are calculated using the same correlations
This model corresponds to low thermal mass collec- as those used for forced convection. It was shown that,
tors. Another model has been developed to account for a laminar flow, the correlation used for forced con-
for thermal mass, e.g., in the case of a Trombe wall: vection is valid for natural convection (Brinkworth
the wall surface temperature is calculated by the thermal et al., 2000).
simulation tool and it is used as a boundary condition Variables m_ and Q_ are inter-dependant, the solution
instead of Tz,adj. is solved by using the Newton method.
The heat transfer calculations presented above are
valid for opaque and semi-transparent PV covers (con- 3.4. PV collector/building interaction
sisting of PV cells laminated in double glazing). The
two resistances R1 and R2 are function of the type of During the simulation process, parameters are ex-
encapsulation of the PV cells, which can be placed on changed at each time step between objects (PV systems
the external or internal pane of the double glazing in our case) and the core of the program. For example,
(Fig. 7), depending on how the cell efficiency decreases as the electrical power needs schedule is defined within
at a higher temperature. the building project, values of the electrical needs are
In case of natural ventilation, several studies con- transmitted to the ÔPV systemÕ object at each time step.
cerning the air flow in an air gap heated by a PV collec- A PV/air collector component interacts with the
tor have been carried out (Brinkworth et al., 2000; adjacent thermal zone(s). During the simulation process,
Moshfegh and Sandberg, 1998). A rather simple method these interactions are taken into account at each time
avoiding to use CFD type calculations (computational step using an iterative algorithm (Fig. 8).
fluid dynamics) allows to predict the air mass flow rate At each iteration i, the heat produced by the PV/air
_ a one-dimensional loop analysis in which the buoy-
m: collector component is calculated, and subsequent zones
70 A. Guiavarch, B. Peuportier / Solar Energy 80 (2006) 65–77
Time step n
component model
T iz
≠T i-1z
≈T i-1z
Time step n+1
Fig. 8. Simplified algorithm for thermal interactions between
the component and the building.
temperatures T iz as well. This loop occurs until the zone
temperatures calculated at a given iteration equal nearly
those calculated at the previous iteration.
3.5. Validation
Concerning the electricity production (see Eq. (1)),
the model has been validated by measurements made
on a stand-alone PV collector by the Centre dÕEnergé-
tique (Ecole des Mines de Paris, Sophia Antipolis)
(Dumbs, 1999). The model, using data provided by man-
ufacturers (Noct, VPM, IPM, etc.), overestimates the elec-
trical yield by roughly 10%.
Concerning the heat production, several models sim-
ilar to the approach presented in Section 3.3 have been
validated (Brinkworth et al., 2000; Moshfegh and Sand-
berg, 1998).
As the study reported in this paper deals with build-
ing integrated photovoltaic collectors, it was necessary
to validate the model of the interaction between the
PV/air collector and the building. Some comparisons
were made between the simulation tool presented in this
paper, and another dynamic simulation tool (developed
by another company) which was also able to simulate
PV/air collectors integrated in buildings. The results
were in good agreement (Guiavarch, 2004).
3.6. Overall productivity and efficiency
If heat is recovered from a PV collector, the perfor-
mance of this system has to be evaluated accounting
both the electricity and the heat produced. But it might
not be appropriate to just sum up both energy values:
the production of 1 kW h PV electricity generally avoids
more primary energy consumption than the production
of 1 kW h heat. This influences the overall optimisation
of the system.
The Eco-invent database edited by the Swiss center
for life cycle inventories in Dübendorf (Frischknecht
and Jungbluth, 2003) provides data regarding the pri-
mary energy needed to produce 1 kW h electricity using
different techniques, and 1 kW h useful heat by various
boiler types. Considering the European mix for the pro-
duction of electricity (37% nuclear, 28% coal, 15%
hydroelectricity, 10% gas, and 10% fuel thermal plants),
and a condensing gas boiler for space heating, the pri-
mary energy corresponding to 1 kW h electricity and
heat can be derived. Using this data, 1 kW h electricity
corresponds to 3.1 kW h primary energy and 1 kW h
heat to 1.3 kW h primary energy. The ratio Cf between
electricity and heat is derived: 3.1/1.3 = 2.44.
To calculate the overall productivity, it is also impor-
tant to take into account the useful heat (the portion of
the produced heat which is useful to reduce the heating
load). The dynamic simulation tool presented in this pa-
per calculates this useful heat, accounting for the ther-
mal interactions between the collector and the
building, by subtracting the heating load of the building
with the solar collector from the heating load of the
same building without the solar collector.
One point to be mentioned concerns the comparison
between the opaque and the semi-transparent collectors:
in the case of a semi-transparent module, the cell area is
smaller than in the case of an opaque module. The over-
all efficiency is related to the cell area (for both electricity
and useful heat) so that opaque and semi-transparent
modules are evaluated in a homogeneous way.
The overall productivity Eov is calculated with the
following expression:
Eov ¼ Epv þ Eth =C f ; ð14Þ
where Epv and Eth are the electricity and the useful heat
produced per unit area of PV cells, and Cf the primary
energy conversion factor.
The overall efficiency is deduced by dividing the over-
all productivity of the collector by the total incident irra-
diation H on the cells:
1
gov ¼ Eov . ð15Þ
H
4. Case studies
Two different typical buildings were selected: a social
apartment building and a single family house located in
France. Concerning the PV technology, this paper con-
centrates mainly on silicon multi-crystalline photovol-
taic cells, referred as ‘‘mc-PV’’. Some simulations have
also been performed with amorphous silicon modules
(only opaque). The amorphous technology is referred
as ‘‘a-PV’’.
 A. Guiavarch, B. Peuportier / Solar Energy 80 (2006) 65–77 71

4.1. Module characteristics

The modules can be either opaque, or semi-

transparent:

• the electric characteristics of the opaque modules are

given by the manufacturer (cf. Table 1). The number
of cells constituting the module is 72 (87% of the
module area is covered by PV cells), and the peak
power is 230 Wp. The thermal resistance of the front
cover of the module (R1), made of hardened glass, is
0.004 W
1 m2 K, and the same value is considered
for the back cover (R2) which is made of different lay-
ers of plastic material (Tedlar);
• the characteristics of the semi-transparent modules
are deduced from those of the PV cells. The number
of cells constituting the module is 62 (75% of the
module area is covered by PV cells), and the peak
power is 198 Wp. The cells are placed at the external
pane of a low emissivity double glazing (Fig. 7), with
a heat losses coefficient U = 1.8 W m
2 K
1. If the
glass placed in front of the PV cells is 4 mm thick,
and assuming that the thermal resistance of the PV
cells is negligible, we can deduce that R1 =
0.004 W
1 m2 K and R2 = 0.383 W
1 m2 K. The
absorption coefficient of the glazing (without PV
cells) av is 0.24, and the transmittance sv is 0.62.
The transparency of the PV collector n, given by
the ratio of the area not covered by the PV cells, is
0.25. The absorber placed on the insulation is made
of black painted steel, and its absorption coefficient
aa is 0.90.
4.2. Social housing
The first building is a social residential building con-
stituted of 52 dwelling units, and built in 1969 (Fig. 9).
Fig. 9. Social residential building (after renovation).
The building has recently been renovated (external insu-
lation on the facades and renewal of windows).
The building has a south oriented facade and four
inhabited storeys. It has been modelled as two different
thermal zones: one thermal zone for the southern part
(zone ‘‘south’’) with a volume of 2666 m3, and one zone
for the northern part (zone ‘‘north’’) with a volume of
3460 m3. The partition between both zones, and the
intermediate floors are 20 cm thick and made of con-
crete. The staircases have been neglected.
The building is 67 m long, 13 m wide, each storey is
2.7 m high, and the total inhabited surface is 3580 m2.
The concrete facade is 20 cm thick, and 8 cm of rock
wool were added on the outer side. The rock wool is pro-
tected by a wood-boarding. The air gap between the
wood-boarding and the insulation is ventilated. The
glazed area of the southern façade is 40% of the whole
surface. The heating set point temperature is 21 C,
and the air ventilation flow rate (including infiltration)
is 0.6 air change per hour.

Table 1
Characteristics of the opaque mc-PV and a-PV module (power specifications at 1000 W/m2, 25 C, AM 1.5)
Parameter mc-PV a-PV
Maximum power, PM 230 Wp 124 Wp
Voltage at maximum power, VPM 34.9 V 30.0 V
Current at maximum power, IPM 6.6 A 4.1 A
Open circuit voltage, VOC 43.6 V 42.0 V
Short circuit current, ISC 7.2 A 5.1 A
Variation coefficient of the open circuit voltage according to the temperature, lVOC
158 mV/C
160 mV/C
Variation coefficient of the short circuit current according to the temperature, lISC +2.085 mA/C +5.1 mA/C
Number of cells in series 72 –
Noct 45 C 46 C
Height, La 1.8 m 2.5 m
Width, w a 0.9 m 0.9 m
Transmission factor of the PV module cover, spva 0.9 0.9
Absorption coefficient of PV cells, apva 0.9 0.9
a
Default value (parameters which are not given by the manufacturer).
72 A. Guiavarch, B. Peuportier / Solar Energy 80 (2006) 65–77
Five different types of integration of the PV collector
are compared. The first type consists in placing the PV
collector over the terrace roof, with PV modules fixed
by metallic frames. Usually, the inclination of the PV
collector is optimised according to the latitude of the
location, but a vertical plane is considered here: the
objective is in a first step to compare the different types
of integration for the same solar radiation, in order to
study how the air temperature, related to the type of
integration, influences the performance of the system.
The thermal interaction between the PV collector upon
the terrace roof and the building is negligible, therefore
this configuration will be called ‘‘non-integrated’’.
The PV collector can be integrated on the façade, un-
der the windows for instance (see Fig. 10).
Several configurations can be distinguished, accord-
ing to the air flow along the PV collector (see Table 2):
• there may be an air gap between the PV collector and
the facade. The air gap is naturally ventilated, espe-
cially by convection due to the heat produced by
the PV collector, this integration type will be called
‘‘ventilated air gap’’;
Fig. 10. PV collector integrated on the façade, under the
windows (partial front view).
Table 2
The different configurations for the integration of the PV
collector
Non-integrated Ventilated air gap
NORTH
NORTH
SOUTH
SOUTH
NORTH
SOUTH
Preheating Preheating ST preheating
south north south
• if the PV collector is directly fixed on the building
wall, this case is called ‘‘without air gap’’;
• in the last case, the fresh air coming from outside
flows through the air gap, is pre-heated by the PV/
air collector, and flows into the building during win-
ter. The air is assumed to be extracted from the build-
ing by a mechanical ventilation system, and the
convection in the air gap is forced. The mechanical
ventilation system and the associated fan power are
assumed to be the same as for the other configura-
tions. In summer, the air vent between the gap and
the building is closed, the air flows outside the gap
through another vent by natural convection, and
fresh air flows directly from the outside into the
building for ventilation purposes.
As described previously, the building is divided in
two thermal zones. If the PV/air collector preheats the
ventilation air of the zone ‘‘south’’, we name the config-
uration ‘‘preheating south’’, otherwise ‘‘preheating
north’’.
Finally, semi-transparent PV collectors are studied.
In this case, an absorber can be added in front of the
insulation of the building wall (see Fig. 11). Like for
the opaque PV collector, the PV/air collector can pre-
heat the zone ‘‘south’’, which is named ‘‘ST and preheat-
ing south’’, or ‘‘ST and preheating north’’ otherwise.
The characteristics of the mc-PV (multi-crystalline)
opaque and transparent modules are reported in Section
4.1.
The opaque mc-PV collector consists of 10 · 10 PV
modules: 10 groups of 10 modules which are connected
in series and connected to an inverter. The total nominal
power is 23,000 Wp (19,800 Wp for the semi-transparent
collector). Each group is connected to an inverter with a
Without air gap
SOUTH
NORTH
ST preheating Fig. 11. PV collector integrated in the facade, side view (semi-
north transparent PV collector preheating the ventilation air; see
Fig. 7 for a detailed side view).
 A. Guiavarch, B. Peuportier / Solar Energy 80 (2006) 65–77
nominal power of 2300 W. As the aim of this paper is to
study the efficiency of the collector only, the characteris-
tics of the inverter are assumed similar in all cases.
The amorphous a-PV collector consists of 8 · 10 PV
modules: eight groups of 10 modules which are con-
nected in series and connected to an inverter. The total
nominal power is 9920 Wp.
The thickness of the air gap is 10 cm (apart if the PV
collector is fixed directly on the facade). The height of
the PV collector is 1.80 m (in the case of a-PV, we use
two widths of the module to get the same height). The
total area of the solar collector is 162 m2. In the case
of preheating, the air flow rate in the solar collector
equals the external ventilation flow rate of the building
zone heated by the collector.
4.3. Single family house
The second building is a single family house, which
consists of two floors (see the plan of the ground floor
in Fig. 12).
The house is modelled as three thermal zones:
• zone ‘‘Day’’: living room and kitchen (ground floor),
and mezzanine (first floor), with a volume of 264 m3;
• zone ‘‘Night ground floor’’: bedroom (ground floor),
with a volume of 135 m3;
• zone ‘‘Night first floor’’: three other bedrooms and
bathroom (first floor), with a volume of 62 m3.
The total heated area is 160 m2. The external walls
are made of 20 cm concrete blocks and 10 cm polysty-
rene placed inside. The roof is insulated with 20 cm rock
wool, and its slope is 45.
The heating set point temperature is 20 C during the
day and 17 C during the night. The ventilation flow rate
(including infiltration) is 0.5 air change per hour.
Fig. 12. Plan of the ground floor of the single family house.
73
PV
collector
Fig. 13. PV collector placed on the roof of the single family
house (front view).
Concerning the PV collector, the different types of
integration are exactly the same as those described in
the previous paragraph (see Table 2), but the PV collec-
tor is placed on the roof which is south oriented with a
45 slope (see Fig. 13).
The simulations are run with only multi-crystalline
PV modules, and the characteristics are the same as
those used in the previous paragraph (see Table 1 for
the opaque PV module). For the semi-transparent mod-
ule, the cells are laminated with the same double glazing.
The mc-PV collector consists of two groups of six
modules which are connected in series and connected
to an inverter. The total nominal power is 2760 Wp
(2376 Wp for the semi-transparent collector).
The thickness of the air gap is 10 cm (if any). The col-
lector is 3.6 m high and 5.4 m wide. In the case of pre-
heating, the air flow rate in the collector equals the
external ventilation flow rate of the spaces heated by
the collector. The collector preheats the ventilation air
for either the zone ‘‘Day’’ only, or the whole building.
This leads to two different configurations: ‘‘Preheating
1 zone’’, or ‘‘Preheating 3 zones’’.
4.4. Meteorological data
Two different locations are considered:
• a site situated near Paris, which corresponds to a
mid-European climate (2800 degree-days);
• a site near Nice, i.e., a Mediterranean climate (1200
degree-days).
‘‘Test reference years’’ climate data files were used for
the simulations (Lund, 1985).
5. Results and discussion
As explained in Section 3.6, the annual thermal load
of the buildings without collector is necessary to
74 A. Guiavarch, B. Peuportier / Solar Energy 80 (2006) 65–77

Table 3
Annual thermal load of the buildings without PV collector, and annual incident irradiation on the plane of the PV collector
Social residential building Single family house
(vertical south plane) (45 slope, south)
Location Paris Nice Paris Nice
Thermal load (kW h/year) 334,960 103,680 14,320 5550
Incident irrad. (kW h/year/m2) 794 1087 1145 1600

calculate the useful heat produced by the PV collector.
These values are given in Table 3 as well as the annual
incident irradiation on the collector plane, which is used
to calculate the mean annual overall efficiency.
As described in the previous paragraph, the simula-
tions are performed according to seven different types
of integration, for two different buildings (with two dif-
ferent slopes for the PV collector), and for two different
climates. The results are reported in Tables 4 and 5 (giv-
ing the annual electricity Epv and the annual useful heat
Eth produced by the solar collector with values given per
unit area of PV cells, the overall productivity Eov and
the overall efficiency gov), and illustrated in Fig. 14
(showing a comparison of the overall productivity for
the different configurations). They are given for mc-PV
collectors, the case of a-PV collectors will be presented
after.
In Fig. 14, we can see that the annual overall produc-
tivity depends strongly on the slope, as expected. For in-
stance, according to Tables 4 and 5, for a non-integrated
collector (the case ‘‘non-integrated’’ will be taken as the
reference case in the following), producing only electric-
ity, the productivity is roughly 110 kW h/m2 for a verti-
cal collector in Paris, and around 160 kW h/m2 for a 45
slope, in the same location.
The overall productivity depends also on the climate,
and is higher in Nice than in Paris. The mc-PV collector
produces more electricity in Nice than in Paris. But, in-

Table 4
Annual electricity (Epv), annual useful heat (Eth), and annual overall productivity (Eov) per unit area of PV cells, and annual overall
efficiency (gov) of the mc-PV collector placed on the façade of the social building, according to the type of integration and the climate
Non-integrated Ventilated Without Preheating Preheating ST preheating ST preheating
air gap air gap south north south north
Epv (kW h/m2) Paris 111.4 112.1 104.4 111.2 111.2 110.8 110.8
Nice 153.6 154.8 141.9 153.0 153.0 152.3 152.3
Eth (kW h/m2) Paris 0 0 0 38.3 56.8 73.9 99.2
Nice 0 0 0 20.6 37.6 47.1 73.9
Eov (kW h/m2) Paris 111.4 112.1 104.4 126.9 134.5 141.1 151.5
Nice 153.6 154.8 141.9 161.5 168.4 171.6 182.6
gov (%) Paris 14.0 14.1 13.1 16.0 16.9 17.8 19.1
Nice 14.1 14.2 13.0 14.8 15.5 15.8 16.8

Table 5
Annual electricity (Epv), annual useful heat (Eth), and annual overall productivity (Eov) per unit area of PV cells, and annual overall
efficiency (gov) of the mc-PV collector placed on the roof of the single family house, according to the type of integration and the climate
Non-integrated Ventilated Without Preheating Preheating ST Preheating ST Preheating
air gap air gap 1 zone 3 zones 1 zone 3 zones
Epv (kW h/m2) Paris 159.0 161.4 146.0 160.2 160.2 159.8 159.8
Nice 222.9 225.9 199.8 223.5 223.5 222.9 222.9
Eth (kW h/m2) Paris 0 0 0 45.5 65.6 150.2 169.4
Nice 0 0 0 35.5 49.1 133.0 144.0
Eov (kW h/m2) Paris 159.0 161.4 146.0 178.9 187.1 221.4 229.2
Nice 222.9 225.9 199.8 238.0 243.6 277.4 281.9
gov (%) Paris 13.9 14.1 12.7 15.6 16.3 19.3 20.0
Nice 13.9 14.1 12.5 14.9 15.2 17.3 17.6
 A. Guiavarch, B. Peuportier / Solar Energy 80 (2006) 65–77 75

Fig. 14. Overall productivity of the mc-PV collector integrated in the apartment building (facade), and the family house (sloped roof),
according to the type of integration and the climate. For the preheating, (1) means zone south for the apartment building and 1 zone
for the family house and (2) means zone north for the apartment building and 3 zones for the family house.

versely, according to Tables 4 and 5, the solar collector
produces less useful heat in Nice than in Paris: the col-
lector produces more heat in Nice, but the heating load
is much lower than in Paris (see Table 3), the produced
heat is therefore less useful.
Except if we compare the two cases ‘‘non-integrated’’
and ‘‘ventilated air-gap’’, we can also see that the pro-
ductivity depends on the type of integration.
For the configuration ‘‘without air gap’’, a non-neg-
ligible loss of productivity occurs. For instance, for the
family house located in Nice, the productivity is around
200 kW h/m2, compared to reference case with
223 kW h/m2. This trend shows the effect of the lack
of cooling by convection behind the mc-PV collector.
In the configurations with air preheating, the overall
productivity is significantly improved compared to the
reference case due to heat production. The thermal pro-
ductivity depends on the heat distribution in the build-
ing. For instance, for the social building, the useful
heat is higher if the collector preheats the zone ‘‘north’’
than if it preheats the zone ‘‘south’’. As previously ex-
plained, the useful heat is function of the heating load
of the preheated zone.
The overall productivity is higher for semi-transpar-
ent mc-PV than for opaque mc-PV, the highest figure
being reached for the collector integrated in the family
house in Nice in the configuration ‘‘ST Preheating 3
zones’’ (282 kW h/m2, the highest productivity obtained
in this study). As explained in Section 3.6, the contribu-
tion of electricity in the overall efficiency is calculated
taking the cell ratio into account. Moreover, the useful
heat is higher for semi-transparent collectors, because
the double glazing transmits a part of the incident radi-
ation which is then absorbed and collected. On the other
hand, semi-transparent mc-PV modules produce slightly
less electricity than opaque mc-PV, because their junc-
tion temperature is a bit higher (around 1 C difference
on average) due to a higher air gap temperature.
The overall annual efficiency can also be compared in
Tables 4 and 5. We can see similar trends for both build-
ings. The sensitivity of the efficiency gov according to the
type of building and the climate is much higher for the
‘‘preheating’’ configurations due to an increased thermal
interaction between the PV/air collector and the
building.
The efficiency of mc-PV non-integrated is roughly
14%. The efficiency is slightly lower for the single family
house than for the social building. The total incident
radiation is higher if the collector is 45 inclined than
if it is vertical, leading to an increase of its junction
temperature.
For the social building located in Paris the efficiency
decreases from 14% in the reference case down to 13%
without air gap, due to the lack of cooling.
The results show how the efficiency is improved by
preheating the ventilation air of the building, the effi-
ciency being higher in Paris than in Nice. The efficiency
depends on which zones are preheated by the solar col-
lector. For instance, for the family house located in
Paris, and with opaque mc-PV, we have gov = 15.6%
for the configuration ‘‘Preheating 1 zone’’ and
gov = 16.3% for the configuration ‘‘Preheating 3 zones’’.
In the case of preheating, the overall efficiency is
higher for the semi-transparent than for the opaque col-
lectors, for the same reasons as those mentioned for the
overall productivity. For instance, for the family house,
we have gov = 16.3% for the configuration ‘‘Preheating 3
zone’’ and gov = 20.0% for the configuration ‘‘ST Pre-
heating 3 zones’’ (the highest efficiency obtained in this
study).
76 A. Guiavarch, B. Peuportier / Solar Energy 80 (2006) 65–77
Table 6
Annual electricity (Epv), useful heat (Eth) and overall productivity (Eov) per unit area of PV cells, and annual overall efficiency (gov) of
the a-PV collector placed on the roof of the social building located in Paris, according to the type of integration
Non-integrated Ventilated air gap
Epv (kW h/m2) 39.8 39.8
Eth (kW h/m2) 0 0
Eov (kW h/m2) 39.8 39.8
gov (%) 5.0 5.0
All the results discussed above were obtained with
the multi-crystalline cells. Table 6 gives the results in
the case of amorphous silicon modules. The simulations
were run for the a-PV collector integrated in the social
building in Paris.
Like for mc-PV collectors, the best way to improve
the overall productivity and efficiency is to use the a-
PV collector to preheat the ventilation air. But contrary
to the mc-PV collectors, the electrical efficiency remains
almost the same for all types of integration. It can even
been seen in Table 6 that the PV collector directly fixed
on the building wall without air gap produces slightly
more electricity than the non-integrated collector. The
reason is that the efficiency of a-PV modules increases
if their junction temperature is higher.
This study concerns the productivity and the effi-
ciency of the system, but financial aspects are not ad-
dressed. However, it shall be kept in mind that the
improvement of efficiency may sometimes lead to a high-
er investment cost. For instance the PV collector pre-
heating the social building is more effective if it
preheats the zone ‘‘north’’ than if it preheats the zone
‘‘south’’, which induces a different cost of the ventilation
system (more pipes for instance). Semi-transparent col-
lectors are more efficient than opaque collectors, but
an absorber has to be added. The absorber in this study
hasnÕt any selective coating, and can be made with sim-
ple black painted steel. The additional cost could there-
fore be reasonable compared to the cost of the whole PV
system.
Concerning the environmental aspects, life cycle
assessment (ISO, 1997) allows the environmental bal-
ance of PV systems to be assessed. A database named
Eco-invent has been elaborated by the Swiss center for
life cycle inventories in Dübendorf (Frischknecht and
Jungbluth, 2003). This database includes, for various
processes (e.g., material production, energy processes),
inventories, i.e., quantities of substances taken from
and emitted into the environment. From these values,
we can derive environmental indicators such as the pri-
mary energy consumption and the contribution in global
warming, expressed in kg CO2 equivalent. One system
studied was a PV collector made of crystalline silicon
cells, with a nominal power of 3 kWp, and for different
types of building integration (Jungbluth, 2003).
Without air gap Preheating south Preheating north
40.0 39.9 39.8
0 40.2 59.5
40.0 56.4 64.2
5.0 6.1 6.6
The electricity and possibly heat produced by this
system lead to a reduction of the impacts related to
the operation phase. This reduction can be evaluated
using the same database considering the European mix
for the production of electricity (37% nuclear, 28% coal,
15% hydroelectricity, 10% gas, and 10% fuel thermal
plants), and a gas boiler for space heating. An environ-
mental ‘‘pay back time’’ can be derived, dividing the
impacts corresponding to the fabrication of the system
by the yearly impact reduction.
In the case of the PV collector placed on the family
house located in Paris (configuration ‘‘non-integrated’’),
the primary energy and CO2 pay back times are around
2.5 years. In the same conditions, if the PV collector is
semi-transparent and used to preheat the ventilation
air of the building (configuration ‘‘ST and Preheating
3 zones’’), the corresponding pay back times are reduced
to around 1.5 years. Innovative PV production processes
are more energy efficient and these values are likely to
decrease in a short term.
6. Conclusions
The productivity and the efficiency of mc-PV (multi-
crystalline silicon based PV collectors) depends on the
type of integration. Fixing a PV collector on a facade
without any ventilated air gap reduces both the produc-
tivity and the efficiency of the modules. However, the
loss of efficiency is not dramatic, and this result has to
be complemented with architectural and economical
aspects.
On the other hand, the efficiency can significantly be
improved if a mc-PV collector is coupled to an air collec-
tor. Using semi-transparent mc-PV modules as a cover
for the air collector, the ventilation air of the building
can be pre-heated and the global efficiency (electricity
+ heat) can reach 20%, to be compared to the reference
case (14%).
The study pointed out that coupling of mc-PV with
an air collector is more efficient using semi-transparent
mc-PV, because of lower heat losses through semi-trans-
parent modules (low emissivity double glazing), and be-
cause the transmitted solar radiation is absorbed behind
the modules.
 A. Guiavarch, B. Peuportier / Solar Energy 80 (2006) 65–77
Concerning a-PV (amorphous silicon based PV col-
lectors), the results showed as well that the best way to
improve the overall productivity and efficiency is to
use the a-PV collector to preheat the ventilation air of
the building. Moreover, the electrical efficiency of a-PV
collectors increases with the temperature.
In the cases where the PV collector is also used to
preheat the ventilation air, designing properly the venti-
lation system (e.g., preheating the whole building, sizing
the air collector) improves the performance in terms of,
e.g., efficiency, energy and CO2 pay back time. Design
can be supported using such a dynamic simulation tool
presented in this paper, in which the implemented model
takes into account thermal interactions between the so-
lar collector and the building.
This simulation tool aims at helping engineers and
architects to compare different PV options, and to pre-
cisely know how much energy the PV collector can pro-
duce, and how its efficiency is influenced by the type of
integration.
References
Brinkworth, B.J., Marshall, R.H., Ibarahim, Z., 2000. A
validated model of naturally ventilated PV cladding. Solar
Energy 69, 67–81.
Duffie, J.A., Beckman, W.A., 1991. Solar Engineering of
Thermal Processes, second ed. John Wiley and Sons, New
York.
Dumbs, C., 1999. Development of analysis tools for photovol-
taic-diesel hybrid systems. Ph.D. Thesis, Ecole des Mines de
Paris, Centre dÕénergétique, Sophia Antipolis.
Frischknecht, R., Jungbluth N. (Eds.), in collaboration with
Althaus H.-J., Doka G., Dones R., Hellweg S., Hischier R.,
Nemecek T., Rebitzer G., Spielmann M., 2003. Eco-invent
77
reports, particularly Chapter 6-XII Photovoltaik, Swiss
center for life cycle inventories, Dübendorf, pp. 1–144.
Guiavarch, A., 2004. Etude de lÕamélioration de la qualité
environnementale du bâtiment par intégration de compo-
sants solaires. Ph.D. Thesis, Université de Cergy Pontoise.
IEA (International Energy Agency), 1996. Photovoltaics in
buildings—a design handbook for architects. Sick, F., Erge,
T., (Eds.), James & James, London.
IEA (International Energy Agency), 2000. Photovoltaic power
systems programme, Task 1. Trends in photovoltaic appli-
cation in selected IEA countries between 1992 and 2000.
Available from: <http://www.iea-PVps.org/>.
ISO standard 14040, 1997. Environmental management—life
cycle assessment—principles and framework.
Jones, A.D., Underwood, C.P., 2001. A thermal model for
photovoltaic systems. Solar Energy 70, 349–359.
Jungbluth N., 2003. Photovoltaik, 2003. Eco-invent reports,
Swiss center for life cycle inventories, Uster.
Kakac, S., Shah, R.K., Aung, W., 1987. Handbook of Single
Phase Convective Heat Transfer. John Wiley and Sons, New
York.
Kays, W.M., Leung, E.Y., 1963. Heat transfer in annular
passages: hydrodynamically developed turbulent flow with
arbitrarily prescribed heat flux. Int. J. Heat Mass Transfer
6, 537–557.
Lund H., 1985. Short reference years and test reference years
for EEC countries. EEC Contract ESF-029-DK.
Moshfegh, B., Sandberg, M., 1998. Flow and heat transfer in
the air gap behind photovoltaic panels. Renew. Sust. Energy
Rev. 2, 287–301.
Peuportier, B., Blanc Sommereux, I., 1990. Simulation tool
with its expert interface for the thermal design of multizone
buildings. Int. J. Solar Energy 8, 109–120.
Ricaud, A., 1997. Photopiles solaires. Presses Polytechniques et
Universitaires Romandes, Lausanne.
Zondag, H.A., De Vries, D.W., Van Helden, W.G.J., Van
Zolingen, R.J.C., Van Steenhoven, A.A., 2002. The thermal
and electrical yield of a PV-thermal collector. Solar Energy
72, 113–128.