Anda di halaman 1dari 9

Exergy analysis of solar collectors, from incident radiation to dissipation

Michel Pons
*
CNRS-LIMSI, BP 133, Rue John von Neumann, Bt. 508, 91403 Orsay Cedex, France
a r t i c l e i n f o
Article history:
Received 20 July 2011
Accepted 22 March 2012
Available online 27 April 2012
Keywords:
Second-law
Dissipation
Direct radiation
Diffuse radiation
Climate
Systematic irreversibility
a b s t r a c t
It is essential to know the actual exergy input of solar radiation in a given location in order to establish
the exergy budget of a solar collector or any other solar-powered process. To do so, the theories on the
entropy of attenuated radiation must be re-interpreted before developing a method for evaluating the
exergy ux from meteorological data. It then becomes possible to build a generic framework for
describing the exergy budget of solar collectors. Three main types of exergy losses can be identied in
this way. The rst is related only to the type of technology chosen for the collector: at-type collectors
and highly concentrating collectors do not have same exergy losses. The second type of exergy loss is
related mainly to heat dissipation, showing that all dissipated heat uxes can be combined as the overall
exergy loss. The third type is related to the utility furnished by the collector. Graphical examples are
shown in a diagram that provides more information than the Sankey diagram.
2012 Elsevier Ltd. All rights reserved.
1. Introduction
Solar energy aims at reducing the use of fossil fuels, whose price
will increase dramatically at some point in time. Moreover, solar
energy is designed to have a lowimpact on global warming. As with
any other developing technology, solar processes will eventually
need to be optimized. Second-lawanalysis, based either on entropy
or on exergy, is a very powerful tool for optimizing processes [1].
The fundamental laws of thermodynamics give the upper bound of
efciency to any process that can be idealized as reversible. This
upper bound is often called the Carnot efciency. Moreover, these
laws show how irreversibilities, which are always present in real
processes, forcibly reduce efciency in comparison with the
reversible reference process. This reversible reference process
allows for a state function called exergy to be dened. Exergy is the
amount of mechanical work that can be produced from a given
power input using a reversible process. From the same input, any
real process will produce less work. With regard to this dened
framework (referred to as the process framework below), the fact
that processes are powered by radiation is a specicity of solar
energy, because radiation and conduction are different by nature.
Indeed, Planck established that blackbody emissions do not trans-
port the same amount of entropy as conduction [2], a distinction
which also applies to exergy. This distinction between radiation
and conduction raises two issues that are specic to solar processes.
First, what is the exergy content of the incident solar radiation in
a given location? Second, given that solar collectors dissipate some
of the incident energy radiatively (through reection and emis-
sion), what is the impact of this distinction on the amount of exergy
lost in radiative dissipation? These two issues are addressed in this
study in order to build a general framework for exergy analyses of
solar-powered processes. This method is applied to solar collectors,
which may be regarded as the most basic solar processes.
Petela was the rst author to derive the equation for the exergy
of blackbody radiation emitted at the Suns temperature when the
heat sink is a blackbody at the environment temperature T
0
[3,4].
Many authors followed Petelas approach [5e10], while a debate
opened up between other authors [11,12] although the disputed
term was very small, as mentioned by Gribik and Osterle. This
debate was re-opened recently [13], and the details of these works
are being reviewed [14]. Terrestrial solar radiation has been
considered as non-blackbody radiation by only a very few authors
[9,15]. This paper follows in the latter authors footsteps in order to
take into account the fact that diffuse and direct solar radiation are
attenuated very differently, a feature that cannot be ignored when
attempting to correlate incident solar exergy with local climates.
The exergetic efciency of a solar-powered system is likely to
depend on the location where the system is operated and, in
particular, on the local distribution between diffuse and direct
radiation. Attenuation of solar radiation by the atmosphere must be
taken into account when comparing the exergetic potential of solar
energy in various places, or when evaluating the exergetic potential
of different energy sources (wind, sun, etc.) in a given location.
Wright [16] developed a theory for evaluating entropy uxes
related to the radiative heat dissipation of solar collectors.
* Tel.: 33 169 858 075; fax: 33 169 858 088.
E-mail address: michel.pons@limsi.fr.
Contents lists available at SciVerse ScienceDirect
Renewable Energy
j ournal homepage: www. el sevi er. com/ l ocat e/ renene
0960-1481/$ e see front matter 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.renene.2012.03.040
Renewable Energy 47 (2012) 194e202
According to this theory, the amount of entropy transported by the
dissipated radiation depends on the spectrum overlaps of three
radiative uxes: 1) short wavelengths reected from incident solar
ux, 2) long wavelengths emitted at a temperature of around
100

C, and 3) long wavelengths emitted at ambient temperature T


0
and reected from environmental emission. Can the fact that some
of this radiation originates from, or eventually turns into, molecular
motion be excluded from this analysis?
The purpose of this paper is to bridge thegapbetweenthe entropy
analysis of radiation alone and the common framework for process
analysis, where many types of energy transfer coexist (mechanical,
radiative, conductive and convective). Solar processes are good
examples of such coexistence. This framework developed for exergy
analysis is thenappliedtoreal meteorological data anddemonstrated
with the examples of at-plate collectors and parabolic concentra-
tors. The causes of exergy losses can then be compared.
2. The thermodynamic framework for exergy analysis of
processes
Numerous textbooks have been dedicated to the concepts of
energy, entropy and exergy [1,17e20]. For the purpose of this study,
a simple generic process is considered. Assuming it is closed and
operated periodically (a steady state is a special case of periodicity),
this process consumes somepower energy, E
p
, toproducesomeuseful
energy, E
u
, while releasing heat Q
e
to the environment, i.e. to external
air, at T
0
. The power energy E
p
can be either mechanical work or heat
supplied at prescribed temperature T
p
(in this paper, heat sources are
thermostats). The utility E
u
can be either mechanical work or heat
exchanged at prescribed temperature T
u
; in the latter case, the heat
ux can either be extracted (E
u
< 0), like in refrigeration or air-
conditioning, or be delivered, like in heat pumping. Mechanical
work may be directlyconvertedinto electricity, or vice versa. The rst
law balance is written as: E
p
E
u
Q
e
0.
To each energy E
p
, E
u
, and Q
e
, is attached an entropy ux S
p
, S
u
,
and S
e
, respectively. Since the nal heat sink is air at ambient
temperature, the result is clearly: S
e
Q
e
/T
0
. For S
x
, with x p or u,
the result is either S
x
0, when the corresponding energy E
x
is work
or electricity, or S
x
E
x
/T
x
, when the corresponding energy is heat.
Note that in all cases the ratio S
x
/E
x
is constant (either zero or 1/T
x
)
and depends only on the nature of the corresponding energy
source. The entropy balance is written as: S
p
S
u
Q
e
/T
0
P
S
0.
By linearly combining the energy and entropy balances according
to E T
0
$S, Q
e
is eliminated from the resulting balance. Another
state function appears B E T
0
$S, called exergy, and which is
proved to be the maximal amount of work that can be produced
from the amount of energy E characterized by its entropy S:
_
E
p
T
0
,S
p
_
E
u
T
0
,S
u
T
0
,

S
B
p
B
u

B
0 (1)
Like entropy, exergy is not conservative. The source term
P
B
T
0
$P
S
is the amount of exergy lost (or destroyed) during
irreversible processes. It is recalled that energy and entropy are
combined according to (E T
0
$S) because S
e
and Q
e
fulll the
general relation for conductive or convective heat exchanges with
external air: S
e
Q
e
/T
0
. A different relation between S
e
and Q
e
could
lead to another combination between E and S, and possibly to
another denition of exergy.
The ratios B
x
/E
x
(1 T
0
$S
x
/E
x
) depend only on the energy
source. They are called exergy factors and denoted by as h below.
They yield the amount of exergy contained in, or transported by,
a given energy, and depend only on the nature of the energy
considered. When applied to a thermostat at temperature T, the
exergy factor is the usual Carnot factor (1 T
0
/T); for mechanical
work, it equals 1. The relation between process efciency (E
u
/E
p
)
and exergy losses can now be derived easily from Equation (1):
E
u
E
p

_
1 T
0
,S
p
=E
p
1 T
0
,S
u
=E
u
_
,
_
1

B
B
p
_

_
h
p
h
u
_
,
_
1

B
B
p
_
(2)
The rst parentheses on the right side contain the ratio of two
exergy factors. It is constant and yields the upper limit for ef-
ciency, usually called the Carnot efciency because it is obtained
through reversible processes (

B
0). The second parentheses
show that efciency is a linear decreasing function of the total
exergy loss (or entropy production) that occurs during the process.
Thus, analyzing exergy losses helps to identify which components
or sub-processes are mainly responsible for losses of efciency. The
latter property paves the way towards optimization.
It should be noted that process optimization, a topic that has
been widely investigated in the literature, usually involves two
steps. The rst step consists of designing the process (i.e. xing all
Nomenclature
b exergy ux density [W m
2
]
c speed of light [2.9979 10
8
m s
1
]
E energy [J]
F
c
collector efciency []
h Plancks constant [6.626 10
34
J s]
i radiative ux density [W m
2
]
I
un
, (J
un
) monochromatic radiation (entropy) luminance
[J m
2
sr
1
(J m
2
sr
1
K
1
)]
j entropyuxdensityassociatedtoradiation[Wm
2
K
1
]
k
B
Boltzmanns constant [1.38 10
e23
J K
1
]
T temperature [K]
Q heat [J]
W mechanical work [J]
S entropy [J K
1
]
X amplication factor of radiative entropy ux due to
attenuation []
B, I, and J respectively have same meanings as b, i, and j, but
integrated on time [J m
2
and J m
2
K
1
]
emissivity []
h exergy factor (exergy/energy) []
n light frequency [s
1
]

B
exergy loss [J]

S
entropy production [J K
1
]
q incidence angle of direct solar radiation [rad]
s StefaneBoltzmann constant [5.67 10
8
W m
2
K
4
]
u
s
solid angle of the Sun [6.79 10
5
sr]
Subscripts
0 ambient air (dead state)
a available
c solar collector
d Dissipation
dr direct
df diffuse
e environment
gl global
p power
s Sun, solar
u utility
M. Pons / Renewable Energy 47 (2012) 194e202 195
the design characteristics) in order to maximize efciency under
the dual constraints of a given utility E
u
and a given investment
cost. The second step consists of xing the investment cost in order
to minimize the total cost, investment plus operation, under the
constraint of a given utility E
u
. Once E
u
is xed, the consumed
power energy is written as: E
p
E
u
$(h
p
/h
u
)
1
(P
B
/h
p
). As
a consequence, the rst optimization step consists of nding,
among all the possible designs with a given investment cost and
a given produced utility, the one design that minimizes the total
exergy loss

B
. Next, the second optimization step consists of
nding, among all the processes optimized according to the
previous criterion, the one process for which any extra investment
made in order to improve efciency (reduce

B
for given utility E
u
)
is compensated exactly by the reduction in operation cost resulting
fromthe decrease in primary energy consumption. This second step
involves nding the best compromise between investment and
operation costs (while expecting that increasing investment should
improve efciency and thus reduce the operation cost). In both
steps, optimization combines economical and technical data. This
leads to a relation between the consumption of power energy and
exergy losses, i.e. phenomena that are clearly localized, clearly
identied and that can be quantied. This is why Equation (2) is
a cornerstone of process optimization. The two quantities of
importance in Equation (2) are the total exergy loss and the exergy
factor of the power source, h
p
. This study applies this framework to
solar collectors and focuses on the exergy factor of terrestrial solar
radiation as a power source, and on the exergy losses of solar
collectors. This methodology can easily be extended to any solar-
powered process.
3. The exergy content of terrestrial solar radiation
3.1. State-of-the-art, from fundaments to solar collectors
Solar radiation incident through the atmosphere is slightly
polarized, but not enough for its exergy content to be modied
[21,22]. Polarization is thus neglected in the following. The mono-
chromatic directional luminance I
un
emitted by a surface at
temperature T is given by I
un

un
,2hc
2
n
3
e
hn=kBT
1
1
; where

un
is the monochromatic, directional emissivity (or attenuation
factor: indeed, non-blackbody emission and attenuation are exactly
equivalent, the factors are thus called emissivity in the following).
Using Plancks theory, the corresponding monochromatic direc-
tional entropy luminance J
un
is given by
J
un

_
2k
B
n
2
=c
2
_
,1 f ,ln1 f f ,lnf ; with
f c
2
I
un
=2hn
3

un
,
_
e
hn=kBT
1
_
1
(3)
For blackbody radiation(
un
1), theintegrationof I
un
andJ
un
over
solid angles and the frequency spectrum yields the emitted uxes
(or more precisely, the total hemispherical ux densities, a point
omitted below) of energy i sT
4
and of entropy j (4/3)$sT
3
(4/
3)$(i/T) [2]. Due to this factor (4/3), blackbody radiation transports
more entropy than the same conductive heat ux. For attenuated
radiation(due to the atmosphere, terrestrial solar radiation is always
attenuated), this factor is evenlarger than (4/3). As withpolarization,
studies [22,23] on spectral selectivity of absorption on the pathway
through the atmosphere show that the effect on the exergy
content of incident solar radiation is negligible. Moreover, whenever
solar processes are selective in regard to frequency, they are
logically designed for less absorbed frequencies. Attenuation is thus
assumed to be isotropic and uniform over the frequency spectrum.
The approximation of isotropic gray body emission (constant
un
) is
very commonly accepted when analyzing solar processes. Within
this framework, the integration of I
un
and J
un
over the hemisphere
(radiative energy and entropy are additive with respect to a solid
angle) andfrequency spectrumyields i sT
4
for the energy ux and
j X
4
3
i
T
(4)
for the entropy ux, where X is a function of only, larger than one
and given by:
X
45
4p
4
_
N
0
x
2
,1 f ,ln1 f f ,lnf dx ; with
f

e
x
1
5
These equations have been presented already in many works
[9,10,12,15,16,21,24,25]. By applying Equations (3)e(5) to emission
at the Suns temperature (T
s
5770 K), some authors have given
numerical approximations of the function X over as wide as
possible ranges of . Landsberg and Tonge [9] gave:
[X() z0.9652 e0.2777$ln() 0.0511$], which is valid for < 0.1.
ztrk et al. [26], and Press [10] gave very similar formula. Wright
[15] gave: [X() z 1 (0.2698 e 0.030$)$ln()], which is valid
within 0.33% for 0.005 < <1; and [X() z1 (0.2647 e0.0173$)$
ln()], which is valid within 0.03% for 0.2 < . None of these works
made an explicit distinction between direct and diffuse radiation,
although this difference is very relevant in the eld of solar energy.
The approach in this study has been adapted to take into account
this feature.
Petela assimilated extra-atmospheric solar radiation with
undiluted blackbody radiation [4] and established the following
formula for the exergy of solar radiation:
b i
_
1
4
3
,
T
0
T
s

1
3
,
T
4
0
T
4
s
_
(6)
At the terrestrial level, solar radiation is always composed of
a direct part plus a diffuse part, both partly diluted, a case to
which Equation (6) should not be applied. The literature on this
subject is somewhat erratic in regard to the exergy input into
solar-powered processes. Most of the time, the question is simply
avoided and analysis does not extend back beyond the net
absorbed heat. Solar exergy is often simply equated with the
radiative ux itself [i.e. state b i], as if radiation were without
entropy. Some studies [27e30], following Jeters work, use the
conduction expression [b i$(1 T
0
/T
s
)], as if there were material
contact between the Sun and the Earth. Other authors [31e35],
following [5e10], apply Petelas formula Equation (6) where i is
equated to the global solar ux, just as if it were undiluted. The
direct and diffuse components of solar radiation are rarely taken
into account [36e38]; the diffuse part is treated as diluted
according to Equation (4) with the phenomenological function X
given in [9,10] while the direct part is treated as undiluted
[Equation (6) again]. However, terrestrial direct insolation rarely
exceeds 1000 W m
2
, even with a very clear sky, while it is
1370 W m
2
just outside the Earths atmosphere. Given that direct
radiation is also attenuated, and that its attenuation factor
strongly differs from that of diffuse radiation, it becomes neces-
sary to develop a new approach.
3.2. Exergy of terrestrial solar radiation for solar collectors or
processes
Within the framework of process analysis, the nal state of
energy is molecular motion in external air at ambient temperature,
M. Pons / Renewable Energy 47 (2012) 194e202 196
i.e. conductive heat, and exergy is unambiguously dened as
B E T
0
$S. For radiation, this combination is written as:
b i T
0
$j. This denition of exergy is used in this paper; it should
be noted that it ensures that exergy cancels out as soon as the solar
ux vanishes (at night), which is a necessary feature of the process
framework.
The distinction between the direct and diffuse components of
incident solar radiation is fundamental in solar energy engineering.
Solar techniques are more or less sensitive to the proportions of
diffuse and direct radiation, a proportion which depends on local
climates (this proportion will differ near coastal regions, in the
Tropics and in deserts). Thus, within the framework of process
analysis and optimization, it is important that the exergy content of
solar radiation depends not only on the intensity of solar ux but
also on the distribution (diffuse vs. direct). Such a denition of solar
exergy makes it possible to:
- quantify the energetic potential of solar energy in a given place,
i.e. in a given climate
- compare this potential to that of other renewable energies, e.g.
wind energy
- compare the exergetic efciency of solar-powered systems
with same utility (e.g. air-conditioning, like in the Orasol
project [39]) but tested in different places, without this
comparison being biased by climatic differences
- rigorously optimize a solar-powered system on the basis of
exergetic analysis, which requires the exergy factor of the solar
ux to be correctly evaluated (see Section 2).
The evaluation of incident exergy must account for the inci-
dence angle and for the uctuations of the solar ux. In regard to
the incidence angle, it seems reasonable to use the horizontal
surface as a reference, as this is the most typical case. More
extensive studies should probably address issues such as shading
and the minimum distance between collectors, or the albedo of the
immediate surroundings around the collectors. Fluctuations at very
different timescales (minute, day, month) are specic for solar
energy. Solar processes are never in steady state and the start-up
and shut-down procedures they undergo at least once a day may
signicantly reduce actual performance compared to steady state
evaluations. The circadian period is the shortest period during
which operation can be approximated as periodical, which is
a condition for the validity of the equations developed in Section 2.
This is why instantaneous uxes of energy, entropy and exergy are
integrated from sunset to sunrise each day, leading to daily
quantities:
I
_
itdt ; J
_
jtdt and B
_
btdt I T
0
,J;
(7)
where I and B are in J m
2
and J in J m
2
K
1
. The effects of climate
and seasons are accounted for by considering data over the length
of a whole year. To the authors knowledge, such an approach is not
currently found in the literature.
It is not only uctuations of the solar ux that occur but also
uctuations of the temperature of external air, which is the dead
state for exergy analysis. This issue goes far beyond the scope of this
paper and has been addressed already by Pons [40] with the
conclusion that the temperature T
0
in the denition of exergy was
to be taken as constant. Indeed, a so-called exergy function
dened as i
(t)
T
0(t)
$j
(t)
would not be a state function and would not
even be conserved in reversible processes. For the sake of gener-
ality with regard to the type of utility, the value of T
0
taken as
reference below is the yearly average outdoor air temperature. It is
thus specic to each location.
3.3. Methodology for evaluating solar exergy
At solar energy sites involving at-plate collectors, or PV cells
without concentration, global radiation is often the only recorded
quantity, because it is sufcient for analyzing such facilities. A more
precise description of solar ux consists of measuring both direct
and diffuse components of solar radiation; this is necessary for
analyzing concentrating solar systems. The data (direct and diffuse)
analyzed herein are recorded with an acquisition time step that is
short enough for representing the rapid variations of solar
radiation.
The solar exergy ux is evaluated as follows from the instanta-
neous values of direct (on the horizontal plane) and diffuse ux
densities, i
dr
and i
df
respectively. Table 1 shows the equations
and data which are specic to either direct or diffuse radiation.
Luminance is deduced from the ux density. For the direct
ux, this calculation involves the angle q between incident radiation
and vertical direction, and the Suns solid angle,
u
s
6.79 10
5
sr, which is so small that dispersion around the
mean incidence can be neglected. For the diffuse ux, the solid angle
is approximated as 2p. For both uxes, the reference luminance
is blackbody luminance at the Suns temperature:
sT
4
s
=p 20.2 10
6
W m
2
sr
1
, which corresponds to the extra-
atmospheric ux density of 1370 W m
2
. The eld of solar energy
provides relevant limits to attenuation factors. The lower limit is
xed by uncertainties in radiation measurements and the minimum
ux for operation, 50 W m
2
. The upper limit is given by the
maximal terrestrial ux density: 1100 W m
2
. The ranges investi-
gated in the literature and reported in Section 3.1 differ signicantly
from the ranges of interest shown in Table 1. This is why the integral
of Equation (5) was accurately numerically integrated for each
relevant range and approximated by specic functions X
dr
() and
X
df
(), shown in Table 1. These functions X
dr
and X
df
are thus well-
adapted to terrestrial solar resources. In their respective investi-
gated ranges, they deviate from the numerically integrated values of
X() by less than 5$10
4
for the direct part, and by less than 2$10
4
for the diffuse part. Each entropy ux, direct j
dr
or diffuse j
df
, is
obtained by introducing the corresponding function X and ux
density i into Equation (4). The corresponding exergy ux is given by
b i T
0
$j.
As the respective solid angles are complementary, the direct and
diffuse components of energy, entropy, and exergy, are additive.
The total solar entropy ux is: j
s
j
dr
j
df
, and the total exergy ux
is: b
s
b
dr
b
df
. In the previous, i
dr
, i
df
, q,
dr
,
df
, X
dr
, X
df
, j
dr
, j
df
, b
dr
,
b
df
, and therefore j
s
and b
s
are all time-dependent. The time
Table 1
The different steps for calculating the incident solar exergy uxes, direct and diffuse,
from meteorological data.
Direct component Diffuse component
Incidence solid
angle
u
s
6.79 10
5
sr 2p u
s
z 2p sr
Luminance i
dr
/[u
s
cos(q)] i
df
/p
Attenuation
factor

dr

i
dr
=u
scosq
sT
4
s
=p

df

i
df
=p
sT
4
s
=p
Range of
interest
for
[0.03e0.8] [10
6
10
5
]
Functions X
X
dr

_
0:973 0:275,ln
0:0273,
_
X
df
() 0.9659 e 0.2776$ln
Entropy ux j
dr
X
dr
(
dr
)$4/3$i
dr
/T
s
j
df
X
df
(
df
)$4/3$i
df
/T
s
Exergy ux b
dr
i
dr
T
0
$j
dr
b
df
i
df
T
0
$j
df
M. Pons / Renewable Energy 47 (2012) 194e202 197
integration of j
s
and b
s
from sunset to sunrise yields the total daily
incident entropy J
s
and exergy B
s
.
3.4. Application to real insolation data
The aforementioned treatment and time integration were per-
formed on three sets of solar data giving the global and diffuse solar
uxes incident on the horizontal plane. The rst set of data was
measured in PIMENT (Physique et Ingnierie Mathmatique pour
lnergie et lEnvironnement, ex-LPBS, Univ. La Runion, France),
located in Saint-Pierre de la Runion (lat. 21.3

S). Global (on


a horizontal plane) and diffuse radiations were recorded from Kipp
& Zonen

pyranometers CM11, one of them equipped with an


equatorial ring; the relative error is less than 5%. Raw data with
a time step of 6 s were averaged over 5 min. The second set was
measured over 2005 and 2006 in PROMES-CNRS (Procds Mat-
riaux et nergie Solaire UPR8521, Odeillo, France, lat. 42.5

N).
Direct, global and diffuse radiations were measured with Kipp &
Zonen

heliometers and pyranometers. Raw data with a time step


of 1 s were averaged over 5 min. For the sake of comparison, we also
used a third set of data generated previously by the Meteonorm

software as part of experimental tests on a solar-powered refrig-


erator in Ouagadugu (Burkina-Faso, lat. 12.3

N), see [41]. Their time


step is 60 min, which is probably too coarse.
These sets of insolation data represent three different climates:
humid tropical in Saint-Pierre de la Runion, Mediterranean moun-
tains in Odeillo (at an altitude of 1550 m), and dry tropical in
Ouagadugu. Finally, the yearly average values of the outdoor
temperature during the period of solar data acquisition were 286 K
in Saint-Pierre de la Runion, 273 K in Odeillo, and 288 K in Oua-
gadugu, respectively. They were taken as reference values of T
0
for
the denition of exergy. Numerical calculations show that changes
of 1 K in these average temperatures modify the values of the
exergy factors dened below by approximately 10
3
. The daily
total incident exergy B
s
mainly depends on the daily total insolation
I
s
, which is why the following graphs are presented in terms of daily
averaged exergy factors: h
s
B
s
/I
s
vs. daily insolation I
s
, for the sake
of clarity. Figs. 1e3 show (among other data, which will be dis-
cussed later) the daily averaged exergy factors (one point per day)
related to total insolation, calculated for Saint-Pierre de la Runion,
Odeillo, and Ouagadugu, respectively. It should be noted that on the
ordinate axis, the range [0.5e0.91] is expanded compared to
[0e0.5]. These total exergy factors, displayed by red crosses, range
from0.68 to 0.7 (overcast conditions) to 0.9e0.91 (clear sky). Gribik
and Osterle [12] or Landsberg and Tonge [9] obtained the same
range [0.7e0.93] for theoretical radiation changing from
completely diffuse to undiluted. Contrary to what is usually
thought, the exergy content of non-concentrated solar energy is
strong. For instance, a Carnot factor of 0.8 corresponds to a heat
source of 1500 K. It can also be noted that data on total exergy are
rather scattered when daily insolation is between 10 and
25 MJ m
2
. Indeed, intermediate daily insolations can be observed
with different scenarios of cloud coverage and therefore different
direct/diffuse ratios. As an example, Fig. 4 presents the global and
diffuse solar uxes in Saint-Pierre for two days when insolationwas
practically the same (20.75 MJ m
2
). It was rather cloudy on
December 24th (thin lines), but the sky was clear on September 1st
(bold lines). This qualitative difference in solar input is shown in the
values of the total exergy factor: 0.36 and 0.85 respectively. The
denition of solar exergy used in the present approach incorporates
information on the quality of insolation, whereas this information
is missing from other denitions that use a xed exergy factor.
Fig. 1. Average exergy factors h for each day in Saint-Pierre de la Runion. Total solar
exergy factor h
s
( ), global solar exergy factor h
gl
( ), and direct solar exergy factor
h
dr
( ). Note the two scales on the ordinate axis.
Fig. 2. Average exergy factors h for each day in Odeillo. Same legends as in Fig. 1.
Fig. 3. Average exergy factors h for each day in Ouagadugu. Same legends as in Fig. 1.
M. Pons / Renewable Energy 47 (2012) 194e202 198
4. Exergy budget of solar collectors
Once the incident exergy is clearly dened, different exergy
losses can be identied and analyzed. One of these can be said to be
systematic for a given technology of collectors, at-plate or with
concentration. One advantage of the present analysis is that it
allows evaluating straightforwardly that systematic exergy loss

Bs
from meteorological data. The information is valuable because it
provides access to the notion of maximum available incident exergy,
B
sa
, simply dened as B
sa
B
s


Bs
. The exergy loss

Bs
may be
related to a part of the incident solar energy that could not be
processed at all by the collector, thus giving rise to the notion of
maximum available solar energy, denoted by I
sa
. The second kind of
exergy losses is related to heat dissipation from collector to envi-
ronment; they will be denoted by

Bd
. Finally, the exergy losses due
to irreversibility in the conversion processes from radiation to
utility are denoted by

Bu
. An issue then arises in regard to the
losses due to dissipation: as there is a difference between radiation
and conduction with respect to transmitted exergy, does the
dissipation mode (radiative or conductive/convective) have an
inuence on the corresponding exergy loss?
4.1. Exergy losses related to collector technology and the nature of
incident radiation
It is well-known in the eld of solar energy that at-plate
collectors and concentrators present fundamental differences
with respect to incident solar radiation. With at-plate technology
both direct and diffuse components of the solar ux are collected,
but they are collected similarly, without taking advantage of the
spatial organization of direct radiation. This feature is accounted for
by the concept of global incident radiation, which is the total
insolation treated as if it were completely diffuse (without any
spatial organization). The exergy income corresponding to this
global radiation can be evaluated easily from meteorological data
using the present approach: the total incident radiation i
s
i
dr
i
df
is treated as if it were completely diffuse. The global entropy ux j
gl
is calculated by applying the above-described procedure for diffuse
radiation to i
s
instead of i
df
, which leads to the global exergy ux
b
gl
i
s
T
0
$j
gl
. Time integration over one day yields the daily global
exergy B
gl
, the only exergy that at-plate collectors can process. The
corresponding exergy loss, P
Bs
B
s
B
gl
, occurs in any at-plate
collector. In this case, the result is: I
sa
I
s
and B
sa
B
gl
.
On the other hand, in highly concentrating collectors (parabolic
dishes, heliostat elds, etc.) only the direct solar ux is focused on
the receiver. Within the limits of innite concentration, diffuse
radiation is not processed at all. The maximal amount of exergy that
can be processed is the direct exergy ux b
dr
i
dr
T
0
$j
dr
mentioned in the previous section (B
dr
after time integration). The
corresponding exergy loss P
Bs
B
s
B
dr
is the diffuse exergy B
df
. In
this case, the result is: I
sa
I
dr
and B
sa
B
dr
.
Figs. 1e3 show the exergy factors corresponding to these
amounts of exergy available to at-plate collectors and to concen-
trators, i.e. h
gl
B
gl
/I
s
and h
dr
B
dr
/I
s
, respectively, where they can
be compared to the total incident exergy factor h
s
.
The global exergy factor, shown by black closed triangles, is
fairly stable at around 0.72, which is just slightly above the value of
0.7 mentioned by Press [10]. The exergy loss corresponding to the
difference h
s
h
gl
can be signicant: up to 20% of the total incident
exergy. Such a loss occurs, as expected, in clear sky conditions.
Nevertheless, an exergy factor of 0.71 corresponds to a heat source
at 1000 K, showing that, contrary to what is usually thought, at-
plate collectors theoretically have a good potential of delivering
heat at high temperatures.
The direct exergy factor, shown by open blue diamonds, lies
between zero and 0.85 (Saint-Pierre and Odeillo) or 0.8 (Ouaga-
dugu). Moreover, the scatter is greater than for total exergy. The
exergy lost through high concentration (dissipation of diffuse
insolation) is not negligible, even for days with very clear sky (11%
of the incident exergy), and can be as high as 50% for days with
intermediate insolation.
Some features would also need further investigation: for
instance, the differences between Figs. 1 and 2, which are probably
related to the local climates, or the lower data scatter in Fig. 3
compared to other scatters, which is probably due to the fact that
the solar data for Ouagadugu are averaged over a much larger time
step.
These results show that the denition of solar exergy in this
paper accounts for the quality of the solar resource, as expected
from the notion of exergy, while a constant exergy factor like in
Equation (6) cannot. Moreover, the present denition provides
access to specic exergy losses that are related to the very principle
of solar collectors, with or without concentration. Such irrevers-
ibilities are systemic, as named by Gicquel [42] and commented by
Pons [43]. These irreversibilities are not related to the design of the
collector: they can be evaluated and compared frommeteorological
data only. Press already made such a comparison [10] based on
theoretical values of instantaneous direct and global exergy uxes.
Without taking into account the climate, he stated that more
exergy is available to collectors with high concentration than to
at-plate collectors. After averaging our data over the whole year,
the results of our analysis (presented in Table 2) showthe opposite:
the average direct exergy is less than the global direct exergy in all
cases (62% compared to 72%). This result demonstrates, a posteriori,
the value of analyzing complete data over days and years. The
exergy available to a low-concentration collector (e.g. a concentra-
tion factor between 2 and 10) might be greater than the previous
Fig. 4. Solar uxes in Saint-Pierre on December 24th (thin lines) and September 1st
(bold lines): global (solid lines) and diffuse (dashed lines).
Table 2
Average daily insolation and exergy contents (total, global and direct) in the three
locations investigated herein [all in MJ m
2
]. The corresponding average exergy
factors are given in italic numbers.
Average daily . Insolation Total exergy Global exergy Direct exergy
In Saint-Pierre 19.6 16.8 (0.86) 14.2 (0.72) 12.3 (0.63)
In Odeillo 16.1 13.8 (0.85) 11.8 (0.73) 9.9 (0.62)
In Ouagadugu 21.9 18.4 (0.84) 15.8 (0.72) 12.7 (0.58)
M. Pons / Renewable Energy 47 (2012) 194e202 199
two, but cannot be evaluated without additional data on the design
of the collector.
Finally, the present analysis requires data on the two compo-
nents of solar radiation, both direct and diffuse; global insolation
alone would not be sufcient. However, most solar-powered
systems are equipped with at-plate collectors and with global
pyranometers. The incident global exergy can be evaluated easily,
but the evaluation of total exergy would require reconstructing the
likely distribution between direct and diffuse components.
4.2. Exergy losses through dissipation
It has been shown above that, for a given ux, radiation and
conduction do not transport the same amount of entropy, and
therefore of exergy. Does this difference mean that conductive and
radiative dissipations generate different exergy losses for a given
rate? This difference, which according to Wright [16] should be
accounted for when analyzing solar collectors, is usually ignored in
second-law analyses of processes involving radiative transfers.
There is indeed a fundamental difference between solar radiation
as a source of energy, and radiative heat exchanges within
a process. Solar radiation is a one way radiation transfer, from Sun
to Earth. This short-wavelength radiative energy is eventually
transformed into radiation emitted by the Earth in long wave-
lengths toward the interstellar vacuum without returning to the
Sun. Conversely, radiative exchanges within a process, or between
a process and its environment, are two-way radiation transfers,
where both parts emit, absorb and reect radiation, the net
exchanged energy resulting from the balance of radiative uxes in
both directions. In addition, these exchanges always occur between
matter and through matter: radiation originates from molecular
motion and is eventually transformed into molecular motion.
Within this framework, a net heat transfer of heat q from a body at
T
1
to another body at temperature T
2
(assuming T
1
> T
2
) unam-
biguously generates an entropy production of p
s
q,T
1
2
T
1
1
,
and an exergy loss of p
b
T
0
,q,T
1
2
T
1
1
, independently of
how the total heat ux q distributes between radiation and
conduction. The exergy loss generated by heat transfer only
involves the heat rate and the temperatures of the sources, but not
the transfer mode. Since outdoor air is basically a conductive heat
source that, due to its innite thickness, also acts as a blackbody
with respect to radiation, it can be concluded that the exergy losses
due to heat dissipation from solar collectors to the environment do
not depend on transfer modes (conduction or convection), radiative
exchanges in infrared wavelengths or the reection of incident
visible wavelengths.
This conclusion, when added to knowledge of available solar
exergy, B
sa
, leads to a highly simplifying but correct method for
dening exergy losses related to dissipation. As any transfer mode
yields the same exergy loss, a single exergy loss may represent the
whole dissipation. By performing a global analysis of the collector,
any dissipated heat rate may be interpreted as the loss of the cor-
responding portion (with respect to the available solar energy) of
the available solar exergy, independently of its actual cause, during
the process by which incident radiation is dissipated into heat
released to the environment. Introducing now the efciency of the
collector F
c
, which is equal to E
u
/I
s
(I
sa
Q
d
)/I
s
, our analysis leads
to an expression of the exergy loss due to dissipation,

Bd
, that can
be evaluated using only the quantities introduced in the previous
sections, I
s
, I
sa
and B
sa
, and F
c
:

Bd

Q
d
I
sa
$B
sa

_
1 F
c
$
I
s
I
sa
_
$B
sa
(8)
4.3. Exergy budget of a solar collector seen as a process
The analysis put forth in Section 2 applies to processes in
periodic operation, i.e. on the circadian cycle for solar collectors.
All the quantities of interest in Section 3 and Equation (8) are
now integrated over a whole day, and the correspondence
between Sections 2 and 3, for 1 m
2
of solar collector, is written
as: E
p
I
s
, and B
p
B
s
. The utility E
u
produced by the solar
collector has a given utility factor h
u
(for instance, if the collector
produces heat at 90

C in Saint-Pierre de la Runion, with
T
0
286 K, h
u
equals 0.21). The produced exergy is then
B
u
E
u
$h
u
. The difference (B
s
B
u
) is the total exergy loss of the
collector; the three parts introduced at the beginning of Section 4
can now be dened.
The rst loss of exergy,

Bs
, is due to the type of solar collector,
from at-type to highly concentrating. In principle, this loss can be
evaluated from the meteorological data (see Section 4.1).
The second loss of exergy,

Bd
, is due to dissipation, all causes
combined (see Section 4.2). It is strongly related to the collecting
efciency F
c
.
The third loss of exergy is the one that occurs between the
exergy factor of the available solar exergy and the exergy factor of
utility h
u
. The general expression of the exergy budget of the solar
collector (B
s
P
Bs
P
Bd
P
Bu
B
u
) can be rewritten straight-
forwardly as B
s
P
Bs
B
sa
Q
d
$(B
sa
/I
sa
) P
Bu
B
u
, where P
Bd
has
been replaced by its expression Equation (8). By introducing the
exergy factor of the available solar energy, h
sa
B
sa
/I
sa
, the exergy
budget becomes: I
sa
h
sa
Q
d
$h
sa
P
Bu
E
u
$h
u
, thus leading to the
expression of P
Bu
:

Bu
E
u
,h
sa
h
u
(9)
Note that this derivation implicitly assumes that h
sa
> h
u
(otherwise the exergy loss P
Bu
would be negative), and therefore
cannot be applied to photovoltaic cells. For photovoltaic cells, the
exergy loss through dissipation, P
Bd
, must be dened differently
from Equation (8).
A graphical application of this analysis will now be presented.
Instead of the usual Sankey diagram, a graph is used [exergy factor
vs. energy] in which the exergy uxes and exergy losses are shown
as areas. Contrary to the Sankey diagram, this graph displays the
energy and exergy budgets together, showing whether a given
exergy loss is accompanied by energy dissipation or not. In Fig. 5,
the energies refer to the total incident solar energy averaged over
one year. The graph on the left side (a) shows the exergy budget of
a at-type collector delivering heat at 90

C with a collecting
efciency of 0.4, the other graph (b) is for a parabolic dish (very
high concentration) delivering heat at 500

C with the same col-
lecting efciency. The solar data are from Saint-Pierre de La
Runion. Of course, it can be seen that the two exergetic ef-
ciencies (B
u
/B
s
) are different, but also that the different exergy
losses do not apply to the same quantities of energy in each case.
For instance, the lower dissipation of the parabolic dish is
compensated by a larger systematic solar loss P
Bs
, which also
corresponds to unused incident ux. Such observations would not
be possible with the Sankey diagram. It also can be seen that, if
their heat dissipation were signicantly reduced, at-type collec-
tors would have the potential of delivering heat at much higher
temperatures than the usual level around 100

C. Indeed, the
exergy loss P
Bu
is larger than the delivered exergy B
u
, when both
are applied to the same quantity of energy E
u
. Finally, the present
analysis requires very little information in addition to the meteo-
rological information, and opens a door to the minimization of
exergy loss, as described in Section 2.
M. Pons / Renewable Energy 47 (2012) 194e202 200
5. Conclusion
Evaluating exergy input into solar collectors and, in addition,
into solar-powered processes requires taking into account two
factors: the radiative nature of the solar energy and the actual
distribution between direct and diffuse radiation. This exergy input
can be evaluated accurately from meteorological data. It appears
that global radiation contains more exergy than direct radiation
alone. In addition, a rst type of exergy loss, related to the tech-
nology of solar collectors (at-type or highly concentrating), can be
identied. A second type of exergy loss can be identied as being
due to dissipation. Here, and contrary to the exergy input, there is
no difference between radiation and conduction or convection with
respect to the exergy losses induced by dissipation to the envi-
ronment. The last type of exergy loss can be attributed to the
irreversibilities during conversion from solar radiation to utility.
The exergy budget of solar collectors, e.g. at-type or parabolic
dishes, can be established using only limited data.
In the near future, the present approach should be applied to
photovoltaic cells and to collectors with lowlevels of concentration.
Acknowledgments
This work was carried out as part of the ORASOL project
(#ANR-06-PBAT-009-05), nanced by the French ANR program
PREBAT and coordinated by Dr. F. Lucas (PIMENT, Univ. La Runion,
France). The solar data for Ouagadugu were kindly supplied by
LESBAT (HEIG-VD, Yverdon-les-Bains, Switzerland), those for
Saint-Pierre de la Runion by M. David (PIMENT), and those for
Odeillo by E. Guillot (PROMES) and J.-M. Gineste (now at JRC).
The present work would not exist without their contributions.
References
[1] Bejan A. Entropy generation minimization. Boca Raton, Florida: CRC Press;
1996.
[2] Planck M. The theory of heat radiation. New York: Dover Publications; 1914.
[3] Petela R. Exergy of heat radiation, transactions of the ASME. Journal of Heat
Transfer 1964;86:187e92.
[4] Petela R. Exergy of undiluted thermal radiation. Solar Energy 2003;74:
469e88.
[5] Badescu V. Maximum conversion efciency for the utilization of multiply
scattered solar radiation. Journal of Physics DApplied Physics 1991;24:1882e5.
[6] Badescu V. Letter to the Editor. Solar Energy 2004;76:509e11.
[7] Bejan A. Unication of three different theories concerning the ideal conver-
sion of enclosed radiation. Transactions of the ASME Journal of Solar Energy
Engineering 1987;109:46e51.
[8] Boehm RF. Maximum performance of solar heat engines [solar-thermal power
plants]. Applied Energy 1986;23:281e96.
[9] Landsberg PT, Tonge G. Thermodynamics of the conversion of diluted radia-
tion. Journal of Physics A Mathematical and General 1979;12:551e62.
[10] Press WH. Theoretical maximum for energy from direct and diffuse sunlight.
Nature 1976;264:734e5.
[11] Spanner DC. Introduction to thermodynamics. London: Academic Press; 1964.
[12] Gribik JA, Osterle JF. The second law efciency of solar energy conversion.
Transactions of the ASME Journal of Solar Energy Engineering 1984;106:
16e21.
[13] Candau Y. On the exergy of radiation. Solar Energy 2003;75:241e7.
[14] Agudelo A, Corts C. Thermal radiation and the second law. Energy 2010;35:
679e91.
[15] Wright SE. Comparative analysis of the entropy of radiative heat transfer
and heat conduction. International Journal of Thermodynamics 2007;10:
27e35.
[16] Wright SE, Scott DS, Haddow JB, Rosen MA. On the entropy of radiative heat
transfer in engineering thermodynamics. International Journal of Engineering
Science 2001;39:1691e706.
[17] Prigogine I. Introduction to thermodynamics of irreversible processes. New-
York: John Wiley and Sons, Inc.; 1962.
[18] Szargut J, Morris DR, Steward FR. Exergy analysis of thermal, chemical and
metallurgical processes. New-York: Hemisphere Pub.; 1988.
[19] Bejan A. Advanced engineering thermodynamics. New York: Wiley; 1988.
[20] Borel L. Thermodynamique et nergtique. Lausanne, Switzerland: Presszes
Polytechniques et Universitaires Romandes; 2005.
[21] Kabelac S. Exergy of solar radiation. International Journal of Energy Tech-
nology and Policy 2005;3:115e22.
[22] Chu SX, Liu LH. Analysis of terrestrial solar radiation exergy. Solar Energy
2009;83:1390e404.
[23] Lesins GB. On the relationship between radiative entropy and temperature
distributions. Journal of the Atmospheric Sciences 1990;47:795e803.
[24] Landsberg PT. Thermodynamics with quantum statistical illustrations. New
York: Interscience; 1961.
[25] Wright SE. The Clausius inequality corrected for heat transfer involving
radiation. International Journal of Engineering Science 2007;45:1007e16.
[26] ztrk M, Bezir NC, zek N. Optical, energetic and exergetic analyses of
parabolic trough collectors. Chinese Physics Letters 2007;24:1787e90.
[27] Esen H. Experimental energy and exergy analysis of a double-ow solar air
heater having different obstacles on absorber plates. Building and Environ-
ment 2008;43:1046e54.
[28] Garcia-Rodriguez L, Gomez-Camacho C. Exergy analysis of the SOL-14 plant
(Plataforma Solar de Almeria, Spain). Desalination 2001;137:251e8.
[29] Koroneos C, Spachos T, Moussiopoulos N. Exergy analysis of renewable energy
sources. Renewable Energy 2003;28:295e310.
[30] Saha SK, Mahanta DK. Thermodynamic optimization of solar at-plate
collector. Renewable Energy 2001;23:181e93.
[31] Chaturvedi SK, Chen DT, Kheireddine A. Thermal performance of a variable
capacity direct expansion solar-assisted heat pump. Energy Conversion and
Management 1998;39:181e91.
[32] Hepbasli A. Exergetic modeling and assessment of solar assisted domestic hot
water tank integrated ground-source heat pump systems for residences.
Energy and Buildings 2007;39:1211e7.
[33] Kara O, Ulgen K, Hepbasli A. Exergetic assessment of direct-expansion solar-
assisted heat pump systems: review and modeling. Renewable and Sustain-
able Energy Reviews 2008;12:1383e401.
[34] Karakilcik M, Dincer I. Exergetic performance analysis of a solar pond.
International Journal of Thermal Sciences 2008;47:93e102.
Fig. 5. Two examples of solar collector exergy budgets, both with a collecting efciency of 40%: (a) at-type collector delivering heat at 90

C, (b) parabolic dish delivering heat at
500

C. Energies on the abscissa (referring to the total incident solar energy), exergy factors on the ordinate.
M. Pons / Renewable Energy 47 (2012) 194e202 201
[35] Torchia-Nuez JC, Porta-Gandara MA, Cervantes-de Gortari JG. Exergy
analysis of a passive solar still. Renewable Energy 2008;33:608e16.
[36] Eskin N. Transient performance analysis of cylindrical parabolic concentrating
collectors and comparison with experimental results. Energy Conversion and
Management 1999;40:175e91.
[37] Eskin N. Performance analysis of a solar process heat system. Energy
Conversion and Management 2000;41:1141e54.
[38] You Y, Hu EJ. A medium-temperature solar thermal power system and its
efciency optimisation. Applied Thermal Engineering 2002;22:357e64.
[39] Lucas F, Boudehenn F, Amblard S, Castaing-Lasvignottes J, Pons M, Le
Pierrs N, et al. ORASOL: a French research program for solar cooling process
optimization. In: Collares Pereira M, editor. EUROSUN-2008 1st int. conf. on
solar heating, cooling & buildings. Lisbon, Portugal: SPES; 2008.
pp. Article#351.
[40] Pons M. On the reference state for exergy when ambient temperature
uctuates. International Journal of Thermodynamics 2009;12:113e21.
[41] Buchter F, Dind P, Pons M. An experimental solar-powered adsorptive
refrigerator tested in Burkina-Faso. International Journal of Refrigeration
Revue Internationale du Froid 2003;26:79e86.
[42] Gicquel R. Mthode doptimisation systmique base sur lintgration
thermique par extension de la mthode du pincement. Revue gnrale de
thermique 1995;34:579e607.
[43] Pons M. Irreversibility in energy processes: non-dimensional quantication
and balance. Journal of Non Equilibrium Thermodynamics 2004;29:157e75.
M. Pons / Renewable Energy 47 (2012) 194e202 202

Anda mungkin juga menyukai