You are on page 1of 12

Available online at www.sciencedirect.

com

Solar Energy 85 (2011) 14571468


www.elsevier.com/locate/solener

Ecient single glazed at plate photovoltaicthermal hybrid


collector for domestic hot water system
Patrick Dupeyrat a,b,, Christophe Menezo b,c, Matthias Rommel d, Hans-Martin Henning a
a

Fraunhofer Institute for Solar Energy Systems (ISE), Heidenhofstrasse 2, D-79110 Freiburg, Germany
b
CETHIL, UMR CNRS 5008/INSA Lyon/UCB Lyon 1, France
c
LOCIE FRE CNRS 3220/Universite de Savoie, Savoie Technolac, France
d
SPF, University of Applied Sciences Rapperswil, Switzerland
Received 30 July 2010; received in revised form 17 March 2011; accepted 4 April 2011
Available online 5 May 2011
Communicated by: Associate Editor Brian Norton

Abstract
This paper deals with the design of a single glazed at plate PhotovoltaicThermal (PVT) solar collector. First, the thermal and electrical performances of several single glazed at plate PVT concepts based on water circulation are investigated, using a simple 2D thermal model, then dierent ways of improvement are presented. It mainly consists in focusing on the heat transfer between PV cells and
uid, and also on the optical properties of materials. Thus, the most appropriate concept conguration has been identied and suitable
material properties have been selected. A prototype collector has been designed, built and tested. A high thermal eciency was reached at
zero reduced temperature. For this level of thermal eciency, the corresponding electrical eciency has is lower than eciency of a standard PV panel using the same technology. However, this solar PVT collector is reaching, in these standard conditions, the highest eciency level reported in the literature.
2011 Elsevier Ltd. All rights reserved.
Keywords: Photovoltaic/thermal; Hybrid solar collector; Thermal collector; Photovoltaic

1. Introduction
In the context of greenhouse gas emissions and fossil
and ssile resources depletion, solar energy is one of the
most promising sources of power. The building sector is
the biggest energy consumer before transport and industry
sectors. Therefore, making use of the buildings envelope
(facade and roof) as solar collecting surfaces is a big challenge facing local building needs: heat, electricity and cooling. However, well oriented roof surfaces for solar
applications are limited. Moreover, when conguring a
suitable building envelope, there is in many cases a conict
between an envelopes use for either hot water production
Corresponding author at: Fraunhofer Institute for Solar Energy
Systems (ISE), Heidenhofstrasse 2, D-79110 Freiburg, Germany.
E-mail address: Patrick.dupeyrat@ise.fraunhofer.de (P. Dupeyrat).

0038-092X/$ - see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.solener.2011.04.002

or electricity production. In this context, the concept of a


multifunctional hybrid photovoltaicthermal collector
seems promising. However, such a concept still faces various barriers. Scientic barriers occur because of the complexity of the multi-physical phenomena that are implied
by the energy operating behavior and operating conditions
of such a concept. Technical barriers are related to material
aging under thermal stress aspects.
A photovoltaic/thermal hybrid solar collector (or PVT
collector) is a combination of photovoltaic (PV) panels and
solar thermal components. In fact, a PVT component is
dened as a device using a PV panel or PV cells as a thermal absorber. The aim of these components is to use the
heat generated in the PV panel and therefore a PVT
device generates not only electrical, but also thermal
energy. Two main families of PVT devices can be identied: concentrated PVT collectors (Coventy, 2005; Rosell

1458

P. Dupeyrat et al. / Solar Energy 85 (2011) 14571468

Nomenclature
a
b
e
gelec
gRef
gTh
gth
g0
s
(sa)e
D
Di
F0
G
G(k)
h

solar absorption coecient ()


PV temperature coecient (/C)
emissivity of absorber plate ()
electrical eciency in PV maximal power point
()
reference electrical eciency ()
thermal eciency in PV open-circuit ()
thermal eciency in PV maximal power point (
)
thermal eciency at zero reduced temperature
()
solar transmission of the glass cover ()
eective transmission absorption coecient ()
outer tube diameter (m)
inner tube diameter (m)
collector eciency factor ()
incident global irradiation (W/m2)
incident spectral irradiation (W/(m2 nm))
convection heat transfer coecient between metal tube and uid (W/(m2 K))

et al., 2005) and at plate PVT collectors. As reported by


dierent authors in recent reviews (Charalambous et al.,
2007; Zondag, 2008; Chow, 2010; Ibrahim et al., 2011),
many congurations of at plate PVT collectors have
been developed. They dier from each other according to
whether an additional glass cover is present (Fujisawa
and Tani, 1997; Chow et al., 2009) and the nature of the
heat transfer uid. Several studies have been performed
on water (Tripanagnostopoulos et al., 2002; Zondag
et al., 2003; Chow et al., 2006), air (Sopian et al., 1996;
Hegazy, 2000; Tiwari and Sodha, 2006) and a few on biuid (Assoa et al., 2007). In this work, our eld of interest
is the water type collector. In this area, several studies have
been led on single-covered or non-covered collectors
(Rockendorf et al., 1999; Kalogirou and Tripanagnostopoulos, 2006; Fraisse et al., 2007) or partially PV covered
(Dubey and Tiwari, 2009).
Several criteria guide the design and the research led on
this eld: energy and exergy criteria of the collector
(Fujisawa and Tani, 1997; Chow et al., 2009; Tiwari
et al., 2009) or energy needs required for system couplings.
At the system level, the studied applications have been for
the purpose of drying (Barnwal and Tiwari, 2008),
coupling with heat pump (Bakker et al., 2005) or building
heating (Fraisse et al., 2007) applications. This makes
energy, and moreover exergy theory analysis not so obvious to implement. This remains an open eld of investigation in regard to criteria choice, and the selected scale
(component or system) leads to a non-trivial energy and
moreover exergy theory analysis.
On an energy point of view at the component scale, the
best electricity performance is achieved for non-covered

hPV-Metal

I
k
n
R(k)
rc
Tin
Tout
Tm
Tamb
TRef
TPV
Tstag
t
U
V
W

heat transfer coecient between PV cells and


metal plate (W/(m2 K))
current (A)
thermal conductivity (W/(m K))
refractive index ()
spectral reection ()
PV packing factor ()
inlet water temperature (C)
outlet water temperature (C)
mean water temperature (C)
ambient temperature (C)
PV cell reference temperature (C)
PV cell temperature (C)
stagnation temperature (C)
thickness (m)
collector thermal losses coecient (W/(m2 K))
voltage (V)
distance between two neighbored tubes (m)

hybrid collectors (Fraisse et al., 2007; Chow et al., 2009).


Indeed, these allow the PV cells operating temperature
to be decreased, and then to increase the electricity conversion eciency. Such components are thus dedicated to
water pre-heating systems, heat pump couplings or direct
heating oors. Fujisawa and Tani (1997) performed an
energy and exergy analysis on this type of component.
From the global energy point of view, they concluded that
the best performance for producing energy gains is
achieved for a single-covered collector, whereas the best
conguration on the side of the exergy approach is an
uncovered collector. The way we are developing our
approach is clearly the response to a specic energy need
of existing or future buildings. The big challenge of PVT
collectors is to reach sucient thermal performance in a
single device, and at the same time comparable electrical
performance to those of PV technologies. In the context
of very ecient buildings, domestic hot water needs still
remain the same in absolute terms. In relative terms its
share will tend to be the most important cause of energy
consumption. In this targeted conguration, the water temperature has to reach at least 60 C. As a consequence,
reaching the same energy performance on both PV and
thermal function in the same product and for the same
operating conditions seems, to be unrealistic. However,
the initial developments carried out in this direction and
presented here show progress and aim to demonstrate that
the potential for improvement is still huge. Fig. 1 shows the
sketch of a single glazed at plate PVT collector based on
water cooling. In this collector concept, silicon crystalline
PV cells operate as a thermal absorber. They are separated
from the outside by a glass cover and an air gap. On the

P. Dupeyrat et al. / Solar Energy 85 (2011) 14571468

Fig. 1. Representation of a at plate covered water PVT collector. (a)


Collector glass cover; (b) air layer; (c) PV cells; (d) metal sheet; (e) metal
tubes; (f) thermal insulation; (g) frame.

one hand, the glass cover reduces the performance of the


PV component by both generating optical loss and leading
to an increase of the cell temperature by greenhouse eect.
This conguration induces an electricity conversion
decrease due to the negative thermal dependency coecient
of silicon crystalline PV cells. On the other hand, the glass
cover strongly increases the thermal performance of the
collector, leading to a better overall thermal energy conversion in comparison to a non-covered collector.
Considering this family of at plate, water hybrid PVT
collectors, many numerical investigations have been carried
out on dierent concept details, but only a few have been
materialized in a prototype.
Rockendorf et al. (1999) placed laminated PV cells onto
the surface of an aluminum solar absorber. This absorber
was inserted into a thermal collector. The air gap between
the absorber plate and the glass cover was 30 mm and the
thickness of the thermal insulation (made of mineral wool)
was 50 mm. The thermal eciency curve of this collector
was measured at an irradiance level of 820 W/m2 and an
outside air velocity of 3 m/s. The aperture area of this collector was 2.1 m2, whereas the solar cell active area was
1.81 m2. The thermal eciency at zero reduced temperature
was 63.3%, with an electrical eciency of 10.3%, leading to
an overall eciency of 73.6%.
Tripanagnostopoulos et al. (2002) have also developed
and tested a at plate glazed PVT collector. This collector
was made of a 0.4 m2 PV poly-Si module mechanically
pressed against a copper sheet and tube absorber. The copper sheet thickness was 0.5 mm, and the copper tube inner
diameter was 10 mm and the copper tube outer diameter
was 12 mm. The distance between two neighboring tubes
was 80 mm. The single glass cover was 2 mm thick and
had a global transmission comparable to low iron glazing.
Based on the aperture area, the thermal eciency at zero
reduced temperature was 59.5% while PV function was
operating with an electrical eciency of 10.5%. An overall
eciency of 70% was then reached.
Zondag et al. (2003) built a PVT collector by gluing a
crystalline-Si PV module onto the surface of a standard
thermal absorber. An aluminum oxide epoxy adhesive
was used for this purpose. The PV module was made of
72 poly-Si 10  10 cm PV cells laminated with EVA (EthyleneVinyl Acetate) front layers, low iron glass cover
(3.2 mm) and Tedlar/Al/Tedlar lm at the backing.
The laminated PV eciency was 10.3% under STC

1459

(standard test conditions, meaning a temperature of


25 C and an irradiance of 1000 W/m2 with an air mass
1.5 (AM1.5) spectrum). The collector aperture area was
0.944 m2 and the solar cell area was 0.72 m2, leading to a
packing factor of 0.77 (solar cells active area and the
whole module surface). The thickness of the metal sheet
was 0.2 mm, the distance between two neighboring tubes
was 95 mm, the tube outer and inner diameters were
respectively 10 mm and 8 mm. A 3 mm thick glass was
placed in front of the PVT absorber. According to the
thermal and electrical measurements made on this PVT
collector, the thermal eciency g0 at zero reduced temperature (when the uid inlet temperature is equal to the ambient temperature) was 54% and the electrical eciency was
8.5%. The corresponding overall eciency (electrical plus
thermal) was therefore 62.5%.
Ji et al. (2007) designed a thermo-siphon at plate glazed
PVT collector by gluing a PV module on the surface of a
at-box aluminum absorber using a silicon gel. The PV
module was made of a transparent front layer, EVA
(EthyleneVinyl Acetate) layers, interconnected crystalline
PV cells and a back opaque TPT (TedlarPolyester
Tedlar) layer. The total absorber area was 1.76 m2, whereas
the solar cell active area was only 1.125 m2, leading to a
packing factor of approximately 0.63. The collector cover
was made of a 4 mm low iron glass separated from the
PV module surface by a 25 mm air layer. A 30 mm thick
thermal insulation was used. For a water 89.2 kg/m2 mass
ow, the collector thermal eciency at zero reduced
temperature was measured at 48.7% and the electrical eciency at 10.15%. This corresponds to an overall eciency
of 58.8% based on the absorber area.
In the cases mentioned above, the thermal eciency of
PVT collectors under PV standard operating conditions
is lower than 64% and their electrical eciency less than
12%. Compared to the zero reduced temperature thermal
performance of a conventional glazed solar thermal collector (80%), the lower eciency is mainly due to electricity
conversion of a part of the incoming radiation which is not
converted into heat. The electrical eciency of PVT collectors is also lower compared to pure PV-modules (same
PV technology). This is due to the additional glass cover
in front of the solar cells, which leads to additional optical
losses. In the following section, we investigate dierent
ways of improving both the thermal and the electrical performances of PVT collectors.
2. Analysis of PVT performances
2.1. Introduction
The most basic approach to build a at-plate PVT
absorber is to connect mechanically a standard crystalline
silicon (c-Si) PV module to the top of the absorber of a at
heat exchanger. The PV laminate is either mechanically
pressed to the top of a at heat exchanger (Tripanagnostopoulos et al., 2002; Fraisse et al., 2007) or glued using an

1460

P. Dupeyrat et al. / Solar Energy 85 (2011) 14571468

additional adhesive layer with good thermal properties.


This layer can be aluminum-oxide-lled two-component
epoxy glue (Zondag et al., 2003) or silicon adhesive (Ji
et al., 2007). In this conguration, the thermal resistance
between the PV cell in the module and the absorber plate
was estimated to be around 0.01 (m2 K/W) by van Helden
et al. (2004). Next, the PVT absorber is inserted in the
frame of a standard collector and covered by a glass cover.
To determine the absorption coecient and the emissivity of a typical crystalline Silicon Si PV module (c-Si), we
have carried out reection measurements. The global emissivity of the module was found to be around 0.9, whereas it
is only 0.05 for a selective coated absorber of a thermal collector. The absorption coecient of the module was found
to be around 0.85, whereas it is around 0.95 for a selective
coated absorber of a thermal collector. Based on the equation presented by Due and Beckman (1991), a simple 2D
thermal model was used to simulate the thermal and electrical eciency curves of a typical PVT collector and the
thermal eciency of a typical thermal collector. The equations used in the module are presented in Appendix A.
Fig. 2 shows the thermal eciency curve of a conventional collector with selective coated absorber as well as
the thermal eciency curve of a PVT collector (with
and without PV operation) according to the reduced temperature (dierence between Tm the mean uid temperature
and Tamb, the ambient temperature divided by G, the global
irradiation). The results of Fig. 2 are obtained from the
model presented in Appendix A. Parameter of the collectors and data concerning the operating conditions have
been extracted from typical thermal collectors tested at
Fraunhofer ISE under such standard conditions. The ow
rate is of 72 kg/h/m2, ambient temperature of 30 C, wind
velocity of 3 m/s and a global perpendicular irradiation of
1000 W/m2. For this calculation, a collector area is of 2 m2.

The absorber layout is a typical sheet and tube made of


copper with a sheet thickness of 0.2 mm, a distance
between two neighboring tubes of 100 mm and the tube
diameter 10 mm. The bottom insulation is 5 cm thick and
the edge insulation 2 cm thick, both having a thermal conductivity of 0.035 W/(m K). The distance between the glass
cover (transmission 0.91) and the absorber plate is 4 cm.
For the standard collector with selective coated absorber,
the absorption coecient used in the calculation is 0.95
and the emissivity 0.05. For the PVT collector, the
absorption coecient is 0.85 and the emissivity 0.9. The
ratio between the area covered by solar cell and the whole
absorber area (packing factor) is 0.8 and the cell eciency
at reference conditions (25 C and 1000 W/m2) is 0.13.
According to our calculations, the optical eciency,
which corresponds to minimal thermal losses (mean uid
temperature is equal to the ambient temperature), is higher
for the standard collector than for a PVT collector. This
result was expected, due to the conversion of a part of
the incoming solar radiation into electricity. However,
the calculations show that the PV conversion represents
only small discrepancies between PVT and thermal collector. In fact, other parameters like the quality of the thermal
contact between the PV cells and the absorber plate, the
absorption coecient and the emissivity, impacts the optical eciency the most. Additionally, the thermal losses of
the PVT collector (corresponding to the slope of the
curve) are much higher than for a thermal collector. This
is due to the higher radiation losses caused by the higher
plate emissivity. These calculations show that the thermal
eciency of PVT collector is not only limited by a physical eect (PV conversion) but mainly by improvable technical issues (emissivity, absorption and contact thermal
resistance). In the following section, a parametric study is
presented, describing possible ways to improve both thermal and electrical performance.
2.2. Parametric study
2.2.1. Glass cover transmission
First of all, thermal and electrical eciencies can be
improved by increasing the solar glass transmission. For
most of the solar collectors, the cover is a standard low iron
content solar glass with a global normal transmission coefcient of about 0.91. Because this parameter not only
impacts the thermal performance, but also strongly the
electrical performance of a PVT collector, it is appealing
to use a double sided coated Anti Reection Coated
(ARC) low iron glass cover for PVT collectors with a
transmission coecient higher than 0.94.

Fig. 2. Thermal eciency curve of a standard collector with selective


coated absorber and of a typical PVT collector with PV operating (mpp)
and without PV operating (open-circuit).

2.2.2. Solar cell technology


In the case of conventional c-Si PV modules, the absorption coecient is optimized for the sensitivity range of the
PV conversion (3001100 nm for silicon). However, in the
case of PVT collectors, it is relevant for the thermal function to consider the absorption coecient for the whole

P. Dupeyrat et al. / Solar Energy 85 (2011) 14571468

range of the solar spectrum (3002500 nm). Because of


their homogenous surface texturing, single-crystalline silicon solar cells (sc-Si) have lower reection losses than
poly-crystalline silicon (pc-Si) solar cells and present a
higher absorption property. In order to check the discrepancy between pc-Si and sc-Si, measurements were carried
out on both pc-Si and sc-Si cells using an integration
sphere. The measured absorption coecients are respectively 0.85 and 0.90 whereas the corresponding electrical
eciency was 0.13 and 0.15. Therefore, it may be preferable for both thermal and electrical sides to use sc-Si solar
cells instead of pc-Si solar cells. This could make even more
sense with a specic low-reecting encapsulation in order
to increase the PVT plate absorption coecient.
2.2.3. PV packing factor
In a conventional c-Si PV module, the solar cells do not
cover the whole surface of the module. The ratio of the PV
area, i.e. the part covered with PV cells, to the whole surface area is usually between 0.8 and 0.9. With the exception
of bi-glass modules, the rear sheet is usually made of white
Tedlar, which possesses high reection in order to reect
the light in the non-active area (between and around cells).
This can lead to a slight increase in the electrical eciency.
However, in the case of PVT collectors, this leads to large
reection losses and therefore to a reduction of usable heat.
Therefore, most of the PVT collectors use a solar absorbing rear sheet. The absorption of those rear materials is
usually greater than the absorption of solar cells (0.95
instead of 0.85 for pc-Si solar cells). Fig. 3 presents the electrical and the thermal eciency of a PVT collector calculated for an absorber plate covered by pc-Si solar cells, for
packing factors of 0.65, 0.8 and 1. The results show that the

Fig. 3. Thermal (full line) and electrical (dashed line) eciency curves of a
typical PVT collector (with PV operating) for dierent PV packing
factors as a function of the reduced temperature.

1461

packing factor has an impact not only on the PV eciency


but also on the thermal eciency. A high packing factor
could be protable to PVT collector.
2.2.4. Heat transfer enhancement
The method of bonding the PV cells to the thermal
absorber metal sheet can be improved in order to minimize
the thermal resistance between PV cells and the heat uid.
As described above, for most of the PVT collectors, an
additional adhesive layer between PV module and metal
sheet is used not only in order to insure thermal and
mechanical contact, but also to absorb mechanical stress
under conditions of thermal expansion. However, this leads
to a low heat transfer rate between PV cell and metal sheet.
In this case, the thermal conductance of this layer is estimated to be around 100 W/m2 K (van Helden et al.,
2004). In order to increase the heat transfer between PV
cells and metal sheet, a more advanced technique can be
used for PVT. This technique consists of laminating
together in one step all the components: the transparent
front glazing (not necessarily glass), the encapsulated
material, the PV cells and the absorber (Zondag, 2008;
Dupeyrat et al., 2009). The aim of the single package lamination is to minimize the thermal resistance between the
PV cells and the metal absorber. However, when reducing
layer thickness between PV cells and thermal absorber,
particular care should be taken, to avoid electrical contact
between both, requiring a specic absorber surface treatment. Instead of three dierent layers (i.e. encapsulant,
TPT and adhesive layers) normally used in the case of
the gluing method, only one layer is necessary in the case
of single package lamination that is proposed here. The
thermal conductance of the layer between PV cells and
metal plate is given by:

1
tEVA
hPV -metal
1
k EVA
The specic conductivity of a EVA layer is
kEVA = 0,35 W/m K (van Helden et al., 2004). An EVA
lamination layer thickness of tEVA = 5  104 m is commonly used. This corresponds to a hPV-metal of about
700 W/m2 K. Additionally, at the solar collector scale,
the heat transfer between metal tubes and uid can also
be improved by using an optimized absorber with appropriate water cooling grid repartition. Instead of a conventional copper sheet-and-tubes absorber, we also simulated
an aluminum absorber with a sheet thickness of 1 mm, a
distance between two neighboring tubes of 50 mm and a
tube diameter of 10 mm. Fig. 4 presents the electrical and
the thermal eciency of a PVT collector calculated for
dierent absorber plates covered with pc-Si solar modules
with dierent heat transfer coecients between PV plate
and channel grid. The results show that this higher transfer
not only strongly increases the thermal eciency of the collector, but also slightly increases the PV eciency (by
reducing the temperature gradient between the mean uid

1462

P. Dupeyrat et al. / Solar Energy 85 (2011) 14571468

Fig. 4. Thermal (full line) and electrical (dashed line) eciency curves of a
typical PVT collector (with PV operating) for two dierent bonding
methods, as a function of the reduced temperature.

and the absorber). Therefore, use of such a method could


be very protable if care is taken to electrically insulate
between cell and absorber, assuming reliability is not an
issue.
2.3. PVT collector modeling
The thermal and the electrical eciency curves are calculated for ve dierent cases in order to evaluate the
cumulative eects of improvements. The cases (a), (b), (c)
and (d) correspond respectively to the technology developed by Zondag et al. (2003), Ji et al. (2007), Tripanagnostopoulos et al. (2002) and Rockendorf et al. (1999). The
case (e) corresponds to the improvements presented in
the previous chapter. Details of all cases are presented in
Table 1.
For the cases (a)(e), the packing factor is 0.80 and the
absorption coecient of the remaining non-PV covered
absorber area is 0.9 (black absorbent foil). The PV
temperature coecient and the reference temperature are

0.0045/C and 25 C, respectively. The glass cover and


PV module emissivity was set to 0.88 and the distance of
the air gap between the glass cover and the module was
4 mm. The bottom insulation was 5 cm thick and the edge
insulation 2 cm thick, both having a thermal conductivity
of 0.035 W/(m K). Both thermal and electrical eciency
curves were always calculated for a solar radiation of
1000 W/m2, a wind velocity of 3 m/s, an ambient temperature of 25 C, a collector tilt angle of 45 and a ow rate of
72 kg/m2/h. The simulated pc-Si PV cells, which are
considered for cases (a) and (c), are assumed to have an
electrical eciency of 0.13, a temperature dependency
coecient of 0.0045 K1 and an global absorption
coecient of 0.85. The simulated sc-Si cell for cases (b),
(d) and (e) have an electrical eciency of 0.15 under standard conditions and an absorption coecient of 0.9. The
value used for the heat conductance of the layer between
the solar cell and the metal sheet is 100 W/(m2 K) as suggested by van Helden et al. (2004) for the gluing method
of cases (a) and (b). It is assumed to be 700 W/(m2 K) for
the lamination and the mechanical pressing used in cases
(c)(e). The copper sheet and tube absorber of cases (a),
(c) and (d) have a sheet thickness of 0.2 mm. The distance
between two neighboring tubes is 100 mm and the tube
diameter 10 mm. The aluminum absorber considered in
cases (b) and (e) is composed of a sheet thickness of
1 mm, a distance between two neighboring tubes of
50 mm and a tube diameter of 10 mm. The solar transmittance of the glass cover was 0.91 in the cases (a) to (d) and
0.94 for case (e). Fig. 5 shows the results in terms of
thermal eciency (PV operating) of each case and their
corresponding electrical eciency curves.
At zero reduced temperature, the calculated thermal eciency for the respective cases are: 0.62 (a), 0.64 (b), 0.65
(c), 0.66 (d) and 0.73 (e), with corresponding electrical eciencies of: 0.088 (a), 0.104 (b), 0.090 (c), 0.103 (d) and
0.110 (e). As was expected, the concept of direct lamination
is the most eective in terms of heat transfer between PV
cell and metal plate. The calculations show also that the
lamination of sc-Si PV cells would improve not only the
electrical performance, due to higher reference eciency
of the cells, but also the thermal performance, because of
their higher absorption coecient. Moreover, it is

Table 1
Main characteristics of (a)(d) existing prototypes and (e) possible optimized PVT collector.
(a)
Pc-Si solar cells
Sc-Si solar cells
Sc-Si solar cells with optically optimized encapsulation
Gluing method
Mechanical pressing
Lamination
Sheet and tube absorber
Flat-box aluminum absorber
Optimized aluminum absorber
Low iron glass cover
High transmittance glass cover

(b)

(c)

(d)

(e)

x
x

x
x

x
x

x
x

x
x
x

P. Dupeyrat et al. / Solar Energy 85 (2011) 14571468

1463

3.1. Absorber lamination

and a refractive nFEP index around 1.39. This refractive


index is lower than for glass (nGlass = 1.52) and therefore,
a reduction of the reection losses at the interface PV/air
is expected in comparison with the reection losses of standard PV modules with glass as front layer (Krauter and
Hanitsch, 1996). All of these layers were laminated
together in a vacuum laminator using standard PV lamination conditions in terms of pressure load, vacuum and temperature, in order to obtain a functional PVT laminate.
Fig. 6 gives a view of the ratio between the solar cell active
area and the full absorber surface (packing factor) and
shows the grid repartition. The packing factor of the laminated PVT absorber was only 0.67, which is quite low
compared to the standard value of 0.80 usually found for
sc-Si PV modules. No delamination between PV encapsulant and metal absorber was observed after all tests were
carried out, but of course a thorough durability analysis
is necessary in the future.
In order to estimate the solar absorption coecient of
the PVT absorber, optical reection measurements were
carried out. Small samples corresponding to the stacks
[Polymer/EVA/Cell/EVA/coated aluminum] and [Polymer/EVA/EVA/Coated aluminum] were laminated. The
optical reection was measured on each sample using an
integrating sphere (Ulbricht sphere) and are presented in
Fig. 7. Curves (a) and (b) correspond to the reection measured on [Polymer/EVA/EVA/Black coated aluminum]
and [Polymer/EVA/Cell/EVA/Black coated aluminum]
respectively. Curve (c) corresponds to the reection measurement made on [Glass/EVA/Cell/EVA/TPT] laminated
with the same type of sc-Si cell. Curve (d) corresponds to
the reection measurements made on a selective coated
absorber of a standard thermal collector. All reection
measurement surface areas were 10  12 mm2.
The laminated black absorber has a reection below 4%
for the whole measurement spectra. In contrast to the

A rigid one side at heat exchanger was designed in


order to perform the whole package lamination process.
The channel geometry was designed in order to achieve a
good homogeneous temperature distribution. Based on this
geometry, a rollbond one side at aluminum absorber was
manufactured. The dimensions of the designed heat
exchanger were 570  1100 mm. The thickness of the aluminum was measured to be approximately 1.2 mm. To
ensure electrical isolation of the solar cells from the metal
absorber, a coating was applied on the surface of the aluminum absorber. The same coating serves as the radiation
absorbing surface for those areas which are not covered by
PV cells. Three strings of six pseudo square sc-Si PV cells
(156  156  0.2 mm) were prepared and connected in series. Those interconnected PV cells were then inserted
between two 0.5 mm EVA lms and deposited on the at
surface of the heat exchanger. A 0.13 mm thick Fluorinated Ethylene Propylene (FEP) lm was used as front
layer. This material has a sucient temperature stability
to be used in a vacuum laminator, a good UV resistance

Fig. 6. Representation of the front and rear side of the developed PVT
absorber.

Fig. 5. Simulated thermal eciency curves (full line) under PV operating


conditions and electrical eciency curves (dashed line) for the cases (a)
(e).

especially strengthened if optically optimized materials are


applied on the top surface of the laminate. Finally, the calculations show that the aluminum absorber, which provides the mechanical support of the PV cells, also allows
the thermal performance to be increased.
At this stage of investigation, an experimental PVT collector has been built in order to implement all these
improvements and to test its eciency in comparison with
existing technologies.
3. Experimental work

1464

P. Dupeyrat et al. / Solar Energy 85 (2011) 14571468

glass with an anti-reective coating. The resulting normal


transmission coecient was measured to be 0.94. The thermal insulation on the back of the absorber is 60 mm thick
and the edge insulation is 40 mm thick. The thermal conductivity of the insulation material is 0.035 W/m K). The
distance between glass cover and PVT laminate is 45 mm.

3.3. Thermal and electrical measurements in steady states


conditions

Fig. 7. Reection measurements made on the laminated samples. (a) On


laminated black coating (b) on laminated sc-Si cell, compared with the
reection measurements made on sc-Si cells laminated in a standard glass/
EVA/Tedlar PV module (c) and made on the selective coated absorber (d).

selective coated absorber, it does not show an increase in


reectivity for wavelengths higher than 1400 nm. The ScSi solar cell laminated with the low refractive index front
layer shows a reduction of the reection losses in the range
[3001200] nm in comparison with the standard module.
Assuming that the absorbed light is equal to 1  R(k),
where R(k) is the measured reection as function of the
wavelength k, it can be possible to calculate the solar
absorption coecient, as it is presented in Eq. (2), where
G(k) is the solar spectrum as a function of the wavelength
k.
R
Gk  1  Rk  dk
R
a k
2
Gk  dk
k

The performance of this PVT collector was measured


at the indoor solar simulator test facility of Fraunhofer
ISE. The eciency values presented below are based on
the aperture surface area. Thermal measurements on the
developed PVT collector were carried out according to
EN12975. The global irradiation was kept close to an average value of 960 W/m2 during all measurements. During
the measurements, the ambient temperature varied between
30 and 33 C, an articial wind was applied with a velocity
of 3 m/s (in a parallel direction to the collector front glass),
a water ow rate of 72 kg/h/m2 was used and the collector
tilt angle was 45.
In the rst set of measurements, the interconnected PV
cells were not connected to a maximum power point
(mpp) tracker. As experiments were carried out in opencircuit conditions, and all absorbed solar radiation was
converted into heat. At zero reduced temperature, the
thermal eciency (a0) is around 0.88. A typical value for
a standard at-plate collector is around 0.8. But compared
to a standard at plate collector, our PVT testing collector
uses a highly transparent glass cover and a thermal
absorber with a good optimized channel grid and thus a
high collector eciency factor F0 . In the second set of measurements, the interconnected PV cells were connected to a

For the solar spectrum AM 1.5, the calculated absorption coecient was respectively 0.96 for the laminated
black coating, 0.93 for the laminated sc-Si cells and lower
than 0.90 for the sc-Si cell laminated in a standard module.
For the whole PVT absorber, the coverage of the solar
cells corresponds to 67% of the whole surface area and
therefore 33% are covered only with this black coating,
leading to a whole solar absorption coecient of about
0.94. This value is quite close to the absorption coecient
of a standard thermal collector (0.95).
3.2. Collector description
A single glazed at plate PVT collector was built using
two of these PVT laminates connected hydraulically in
parallel. The outer dimensions of the experimental collector were 1360  1350  120 mm and the absorber area
was 1.27 m2. Both PVT laminates were electrically connected in series. The front cover was 4 mm thick low-iron

Fig. 8. Thermal eciency curve of the developed PVT collector with PV


operating in open-circuit mode (triangles) and in maximal power point
(squares) as a function of the reduced temperature.

P. Dupeyrat et al. / Solar Energy 85 (2011) 14571468

mpp tracker and therefore the incoming radiation that is


converted into electricity was extracted from the collector.
Fig. 8 presents the results of these measurements and the
corresponding thermal eciency curve (polynomial t) in
both oc and mpp modes. In open-circuit, the t coecients
were a0 = 0.88; heat loss coecients a1 = 7.00 (linear part)
and a2 = 0.026 (quadratic part). For the PV operating
at maximal power point, the t coecients were
a0 = 0.79; heat loss coecients a1 = 6.36 (linear part) and
a2 = 0.035 (quadratic part).
IV measurements made with the mpp tracker during
the thermal measurements are detailed in Fig. 9 for ve different mean uid temperature levels. As expected, the uid
temperature has a big inuence on the PV module parameters. In fact, a large increase in the open-circuit voltage
Voc is observed with increasing water mean temperature,
whereas a slight increase of the short circuit current was
observed as well. For each IV measurement, the maximal
power point was extracted and the electrical eciency was
calculated.
The electrical eciency of the collector was calculated
using the IV characteristics presented in Fig. 9. The electrical and the thermal eciencies of the PVT collector in
hybrid mode are plotted in Fig. 10 as functions of the
reduced temperature. This representation is not conventional in regard to the PV eciency. However, it allows
us to evaluate an overall eciency in hybrid operating
mode for a reduced temperature equal to zero. At zero
reduced temperature, the electrical eciency is about
0.087. As previously mentioned, the thermal eciency is
around 0.790, leading to an overall eciency of 0.877.
Based on the thermal model, the thermal and electrical eciencies were also calculated using the collector layout
parameters in order to validate our approach. Those eciency curves were obtained using the model and also pre-

Fig. 9. IV curves measured according to the uid mean temperature


(corresponding to the mpp thermal measurements of Fig. 8).

1465

Fig. 10. Electrical (circles) and thermal (square) eciencies measurements


of the PVT collector in hybrid mode as functions of the reduced
temperature and eciency curves based on the thermal model presented in
Appendix A.

sented in Fig. 10. The simulated curves are in good


agreement with the experiments. The model can therefore
be considered as valid for this case.
3.4. Stagnation temperature
As described by Due and Beckman (1991), the stagnation temperature under high radiation with no uid owing
through the collector is usually much higher than the
ordinary operating temperature range. For domestic hot
water applications, the stagnation conditions can occur
several times a year. This happens mainly during times in
which the storage (water tank) is completely loaded, as well
as some other cases such as power failure, control problems, servicing, summer shutdown, or other causes leading
to no-ow conditions. For conventional thermal collectors
with selective absorber coating, the stagnation temperature
is between 180 200 C for an irradiation of 1000 W/m2
and an ambient temperature of 30 C. However, the thermal eciency curve of a PVT collector shows higher thermal losses and therefore a lower stagnation temperature
can be expected. Measurements are performed according
to the stagnation test in EN12975-2. The corresponding
radiation is 1000 W/m2 radiation, 30 C ambient temperature, a wind velocity <1/s, tilt angle 45, no uid inlet and
outlet connection and with no uid owing through the
collector. In open-circuit mode, the stagnation temperature
Tstag is measured to be around 142.5 C, whereas the maximum power point is reached at only 136.3 C. The dierence between the stagnation temperature reached by the
collector in open-circuit and in mpp mode is due to the
conversion of a part of the incoming radiation into heat,
even at temperatures higher than 135 C.

1466

P. Dupeyrat et al. / Solar Energy 85 (2011) 14571468

In addition to the possibility of damages that might


occur in conventional thermal collectors, exposure to stagnation conditions can rapidly degrade or even boil the heat
transfer uid. However, the boiling point of the ethylene
glycol solution is relatively high (130150 C, depending
on % of glycol) and therefore will not be considered as a
main limiting factor for PVT collectors. In the case of
PVT collectors, the exposure to stagnation conditions
obviously leads to other specic eects related to the fact
that the absorber is a kind of PV module. In fact, the
PVT absorber is made in a lamination process, using
EVA (ethylene vinyl acetate) as encapsulating material.
This is very important for the following investigations, as
this material can be damaged at high temperatures. This
accelerates its aging and may provoke coloration (reduced
absorption) and delamination. It has to be highlighted that
there is no existing certication or test of standard
glass/EVA modules above 85 C and therefore no real
quantitative study of the reliability of glass/EVA modules
above 85 C. Zondag and van Helden (2002) investigated
the stagnation behavior of PVT collectors and showed
that the risk of EVA delamination increases greatly above
135 C. Additionally, they mentioned the yellowing of
EVA material under the combination of high temperature
and UV. However, their system simulation showed that
the stagnation condition in European countries should stay
below 135 C. The other limiting factor is related to the use
of crystalline silicon (c-Si) for PV cells. Indeed, c-Si has a
negative temperature coecient and its eciency drops to
about 0.45% C-1, which means that PV production drops
with higher temperatures (for the same irradiation). It is
therefore desirable to cool the PV cells as eciently as
possible and to privilege PV production during periods
when no hot water is needed. Also, the lifetime of silicon
PV modules is guaranteed only for operating temperatures
below 85 C. For this reason, the upper limit of peak
temperatures must be set at 85 C, but lower temperatures
are desirable during holiday periods to enhance PV
production.

Fig. 11. Thermal eciency curves of a standard thermal solar collector


with selective absorber (square), with non-selective absorber (cross) and
the laminated PVT collector under PV operation (dashed line), measured
under the same conditions.

electric eciency can be achieved by increasing the coverage of the absorber with PV cells. As was presented in
Fig. 3, this will have also an eect on the thermal eciency
and a slight decrease of thermal eciency is expected. The
thermal results of the experimental PVT collector under
PV operation (mpp) are presented in Fig. 11 and compared
to the measured performance characteristic of two conventional solar thermal collectors respectively with and
without a selective coating. The thermal eciency at zero
reduced temperature of the laminated PVT collector
seems to be very close to the eciency of a collector with
and without a selectively coated absorber. The thermal
losses are comparable to these of a collector with nonselective coated absorber and are much higher collector
with selective coated absorber. This is mainly related to
the higher emissivity in the absorber areas covered by PV
cells, which are not spectrally treated with the selective
thermal coating.

3.5. Discussion
4. Conclusion
Single-crystalline silicon cells were laminated with a very
high transmission thin lm on the surface of a coated optimized metal absorber. This PVT absorber was built into a
glazed collector and tested. The measured overall collector
eciency is considerably higher compared to the results
presented previously in the literature and shows that the
developed method may bring signicant improvements in
the manufacturing of PVT collectors. However, compared
to a standard sc-Si module (around 14%), the electrical eciency results obtained in these measurements are still low.
However, the main disadvantages for the electrical eciency is, in our case, the presence of the additional glass
cover, which leads to an additional electrical eciency
reduction (due to additional optical losses), and the low
PV packing factor. However, a signicant increase of

Thermal and electrical performances of several PVT


collector concepts have been evaluated through a 2-dimensional modeling approach. This model provides performance prediction under steady-state conditions. At this
stage of the PVT concept design, this is sucient to identify dierent ways of improvement. The results of these
investigations indicate that the direct lamination of sc-Si
PV cells on an optimized metal heat exchanger leads to
the best results among the investigated concepts. An experimental PVT collector was built using the single package
lamination method, focusing on an improved heat transfer
between PV cells and cooling uid and on improved optical
performance (anti-reective coating on the glass cover).
This collector was tested.

P. Dupeyrat et al. / Solar Energy 85 (2011) 14571468

The thermal eciency at zero reduced temperature was


measured at 79% under PV operation with a corresponding
electrical eciency of 8.8%, leading to a high overall eciency of almost 88%. These experimental results indicate
a signicant improvement of both thermal and electrical
performance in comparison to previous work on PVT collector concepts.
The electric eciency of the experimental PVT collector is also signicantly lower than that of good sc-Si PV
modules using the same solar cell technology; however, this
is mainly due to a low packing factor, and therefore future
work will focus on optimizing the overall geometry of the
device.
The optical performance of the experimental collector
reaches values which are as high as good solar thermal only
collectors. However, the heat loss coecients of this collector are signicantly higher than those of high ecient at
plate collectors, but still in the same range as some less
ecient solar thermal only collectors on the market. In
addition, improvements towards advanced production
technologies and a thorough durability analysis especially
under stagnation conditions will be subject of future
R&D work.
Acknowledgments
The authors of this work would like to thank M. Tranitz
and H. Wirth for their help during the lamination process
of the PVT absorber, P. Hofmann for his support during
the testing of the collector in the solar simulator laboratory, the company Q-Cells for providing the solar cells used
in this collector and G. Kwiatkowski and D. Binesti from
Electricite de France (EDF R&D) for their support during
this project. This work was funded by the department
EnErBat of EDF R&D.
Appendix A
A. 1. PVT collector performances modeling
In order to evaluate the combined eects of these
improvements, a simple 2D thermal model was used to simulate the resulting thermal and electrical eciency curves.
The aim of this model is to evaluate both thermal and electrical eciency curves of a PVT collector for dierent
designs and over the range of possible operation temperatures. The thermal eciency of a standard thermal collector is given by Eq. (3) where DT corresponds to the
dierence between the absorber temperature and the ambient temperature and G corresponds to the incoming solar
radiation (Due and Beckman, 1991)
gTh saeff  U

DT
G

In this equation, the term (sa)e is the eective transmission-absorption product and U the thermal losses coecient of the collector. In the case of a PVT collector, the

1467

solar cells are connected to a MPP-Tracker and produce


electricity. Part of the incoming radiation is converted into
electricity and therefore cannot be transferred into heat,
leading to a modication of the PVT thermal eciency
gTh , see Eq. (4), where gth is the thermal eciency without
extraction of electricity and gelec the corresponding electrical eciency.
gTh gTh  gElec

where gTh is the thermal eciency without extraction of


electricity and gElec the corresponding electrical eciency.
This can be developed according to Eq. (3) as:


DT
G
DT
 sc rc gref 1  bT pv  T ref
U
G

gTh saeff  U 
saeff

where sc is the transmissivity of the glass cover, rc the PV


packing factor, gref the electrical reference eciency, b
the PV conversion temperature dependency coecient,
TPV the PV temperature and Tref the PV reference
temperature.

This impacts the values of saeff and U*. In fact, at a
point, corresponding to a situation when the absorber temperature (assumed to be the PV cells temperature) is equal
to the ambient temperature, the term with the thermal
losses disappears and Eq. (5) becomes
gTh saeff
saeff  sc rc gref 1  b  T amb  T ref

And the thermal losses of the PVT collector U* can be


calculated as presented in Eq. (7).
U  U  sc rc gref bG

Usually, the thermal eciency curves are measured and


calculated according to Eq. (8) where DT corresponds to
the dierence between the mean uid temperature and
the ambient temperature and F0 is the collector eciency
factor.


DT
0
gTh F saeff  U
8
G
The collector eciency factor F0 characterizes the thermal quality of a solar collector (Eisenmann et al., 2004).
As F0 is inuenced by the tube distance W and the absorber
plate thickness tp, F0 is also correlated with the material
content of absorber sheet and tubes. The description of
the factor F0 is given in Eq. (9).
F 0PVT

1=U 
0
U  @DW D

W
p

tanh

U
W D=2
k si tsi k p tp

p

U
W D=2
k si tsi k p tp

1 h

PV -metal

pDWi hfi

 A
9

1468

P. Dupeyrat et al. / Solar Energy 85 (2011) 14571468

As for a standard thermal collector, this collector eciency factor depends on the absorber geometry. kp and
tp are respectively the thermal conductivity and the thickness of the metal sheet. W is the distance between two
neighbored tubes, D is the tube outer diameter and Di is
the inner diameter. h represents the heat transfer coecient between metal tube and uid (convection). In the case
of a PVT collector, the collector eciency factor also
depends on the values of ksi and tsi (respectively the thermal
conductivity and the thickness of the PV material) and on
the heat transfer coecient between PV cells and metal
plate (conduction) hPV-Metal.
From these equations a whole thermal eciency curve
can be obtained. The calculation is grounded on the variation of inlet temperature while the other parameters are
kept constant, using a double loop iteration on the set of
equations. This process is similar to the one described by
Due and Beckman (1991).
References
Assoa, Y.B., Menezo, C., Fraisse, G., Yezou, R., Brau, J., 2007. Study of
a new concept of photovoltaicthermal hybrid collector. Solar Energy
81, 11321143.
Bakker, M., Zondag, H.A., Elswijk, M.J., Strootman, K.J., Jong, M.J.M.,
2005. Performance and costs of a roof-sized PV/thermal array
combined with a ground coupled heat pump. Solar Energy 78, 331
339.
Barnwal, P., Tiwari, G.N., 2008. Grape drying by using hybrid photovoltaicthermal (PV/T) greenhouse dryer: an experimental study. Solar
Energy 82, 11311144.
Charalambous, P.G., Maidment, G.G., Kalogirou, S.A., 2007. Photovoltaic thermal (PV/T) collectors: a review. Applied Thermal Engineering
27, 275286.
Chow, T.T., He, W., Ji, J., 2006. Hybrid photovoltaic-thermosyphon
water heating system for residential application. Solar Energy 80, 298
306.
Chow, T.T., Pei, G., Fong, K.F., Lin, Z., Chan, A.L.S., Ji, J., 2009.
Energy and exergy analysis of photovoltaicthermal collector with and
without glass cover. Applied Energy 86, 310316.
Chow, T.T., 2010. A review on photovoltaic/thermal hybrid solar
technology. Applied Energy 87, 365379.
Coventy, J.S., 2005. Performance of a concentrating photovoltaic/thermal
solar collector. Solar Energy 78, 211222.
Dubey, S., Tiwari, G.N., 2009. Analysis of PV/T at plate water collectors
connected in series. Solar Energy 83, 14851498.
Due, J.A., Beckman, W.A., 1991. Solar Engineering of Thermal
Processes. John Wiley, New York.
Dupeyrat, P., Menezo, C., Hofmann, P., Wirth, H., Kwiatkowski, G.,
Binesti, D., Rommel, M., 2009. Development of a high performance

PVthermal at plate collector. In: Proceedings of CISBAT Conference, Lausanne.


Eisenmann, W., Vajen, K., Ackermann, H., 2004. On the correlations
between collector eciency factor and material content of parallel ow
at-plate solar collectors. Solar Energy 76, 381387.
Fraisse, G., Menezo, C., Johannes, K., 2007. Energy performance of water
hybrid PV/T collectors applied to combisystems of Direct Solar Floor
type. Solar Energy 81, 14261438.
Fujisawa, T., Tani, T., 1997. Annual exergy evaluation on photovoltaic
thermal hybrid collector. Solar Energy Materials and Solar Cells 47,
135148.
Hegazy, A.A., 2000. Comparative study of the performances of four
photovoltaic/thermal solar air collectors. Energy Conversion and
Management 41, 861881.
Ibrahim, A., Othman, M.Y., Ruslan, M.H., Mat, S., Sopian, K., 2011.
Recent advances in at plate photovoltaic/thermal (PV/T) solar
collectors. Renewable and Sustainable Energy Reviews 15, 352365.
Ji, J., Lu, J.P., Chow, T.T., He, W., Pei, G., 2007. A sensitivity study of a
hybrid photovoltaic/thermal water-heating system with natural circulation. Applied Energy 84, 222237.
Kalogirou, S.A., Tripanagnostopoulos, Y., 2006. Hybrid PV/T solar
systems for domestic hot water and electricity production. Energy
Conversion and Management 47, 33683382.
Krauter, S., Hanitsch, R., 1996. Actual optical and thermal performance
of PV-modules. Solar Energy Materials and Solar Cells 4142, 557
574.
Rockendorf, G., Sillmann, R., Podlowski, L., Litzenburger, B., 1999. PVhybrid and thermoelectric collectors. Solar Energy 67, 227237.
Rosell, J.I., Vallverdu, X., Lechon, M.A., Ibanez, M., 2005. Design and
simulation of a low concentrating photovoltaic/thermal system.
Energy Conversion and Management 46, 30343046.
Sopian, K., Liu, H.T., Yigit, K.S., Kakac, S., Veziroglu, T.N., 1996.
Performance analysis of photovoltaic thermal air heaters. Energy
Conversion and Management 37, 16571670.
Tiwari, A., Sodha, M.S., 2006. Performance evaluation of hybrid PV/
thermal water/air heating system: a parametric study. Renewable
Energy 31, 24602474.
Tiwari, A., Dubey, S., Sandhu, G.S., Sodha, M.S., Anwar, S.I., 2009.
Exergy analysis of integrated photovoltaic thermal solar water heater
under constant ow rate and constant collection temperature modes.
Applied Energy 86, 25922597.
Tripanagnostopoulos, Y., Nousia, T., Souliotis, M., Yianoulis, P., 2002.
Hybrid photovoltaic/thermal solar systems. Solar Energy 72, 217234.
van Helden, W., van Zolingen, R., Zondag, H.A., 2004. PV thermal
systems: PV panels supplying renewable electricity and heat. Progress
in Photovoltaics: Research and Applications 12, 415426.
Zondag, H.A., van Helden, W.G.J., 2002. Stagnation temperature in PVT
collectors. Presented at PV in Europe From PV Technology to
Energy Solutions Conference and Exhibition, Rome, Italy.
Zondag, H.A., de Vries, D.W., van Helden, W.G.J., van Zolingen, R.J.C.,
van Steenhoven, A.A., 2003. The yield of dierent combined PV
thermal collector designs. Solar Energy 74, 253269.
Zondag, H.A., 2008. Flat-plate PVthermal collectors and systems: a
review. Renewable and Sustainable Energy Reviews 12, 891959.

You might also like