Upload
l-narvarte
View
266
Download
3
Embed Size (px)
Citation preview
Tracking
PROGRESS IN PHOTOVOLTAICS: RESEARCH AND APPLICATIONS
Prog. Photovolt: Res. Appl. 2008; 16:703–714
Published online 7 August 2
*Correspondence to: L. NarCiudad Universitaria s/n, 28yE-mail: [email protected]
Copyright # 2008 John Wil
008 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/pip.847
Applications
and Ground Cover Ratio
L. Narvarte*,y and E. LorenzoInstituto de Energıa Solar, UPM, Ciudad Universitaria s/n, 28040 Madrid, SpainEnergy yield and occupation of land are two parameters that must be optimized when designing a large PV
plant. This paper presents the results of simulating the energy yield of flat panels for some locations and
different tracking strategies as a function of the ground cover ratio. Some interesting results for design
purposes, such as the optimal solar trackers depending on the land availability or the energy gains of every
tracking strategy, are shown. For example, the energy gains associated to one north–south axis tracking,
referenced to static surfaces, ranges from 18 to 25%, and from 37 to 45% for the two-axes tracker for
reasonable ground cover ratios. To achieve these results, a simulation tool with appropriate models to
calculate the energy yield for different solar trackers has been developed. Copyright # 2008 John Wiley &
Sons, Ltd.
key words: PV plants; grid-connected systems; tracking; ground cover ratio
Received 3 March 2008; Revised 24 June 2008
INTRODUCTION
Solar tracking has been used with PV flat modules for
more than 20 years.1 The available literature describes
several cases with good performance in terms of both
reliability and energy gains.2–15 Typical costs of PV
grid-connected systems at current Spanish market are
around 5s/Wp for static arrays (3s/Wp for PV
modules plus wiring, 0�5s/Wp for support structure
plus civil works, 0�5s/Wp for inverters and 1s/Wp for
others) and 6s/Wp for two-axes tracking (support
structure plus civil works are now 1, 5s/Wp). Then,
investment increase is about 20% while energy yield
increase is about 40%. This is why trackers are
flourishing in this market and also why this can
continue, meanwhile PV modules cost remains larger
than 0�8s/Wp. We estimate that about 70% of the
106MW installed during 2006 are combined with solar
varte, Instituto de Energıa Solar, UPM,040 Madrid, Spain.m.es
ey & Sons, Ltd.
trackers. A recent market survey16 has identified up to
22 companies offering a large variety of trackers.
When many PV generators are concerned, there is an
obvious relationship between land occupation and
energy yield; increasing land occupation reduces
mutual shading and, therefore, increases energy yield.
However, this is not without a price, because land-
related costs (land, wiring, fencing, civil works, etc.)
also increase. Hence, the study into the relationship
between land occupation and energy yield is a key
point in the design of large PV plants. At the IES-UPM,
we are often required to advise on practical questions
about the design of PV tracking plants: What is the
recommended tracker for a given terrain? What is the
recommended separation between trackers? etc. In
fact, answering such questions has been the main
motivation for this work.
This paper deals with the relationship between land
occupation and energy yield of flat panels for different
tracking possibilities: two axes, single vertical axis,
single horizontal axis and single tilted axis. As a matter
of reference, static surfaces are also considered. Thus,
tracking geometry has been reviewed and a simulation
exercise for different places has been carried out.
704 L. NARVARTE AND E. LORENZO
TRACKING GEOMETRY
The receiver position and the incident angle are
relevant parameters as far as tracking is concerned.
Tracking geometry descriptions can be found in the
available literature for many years. Those based on
matrix17 are particularly consistent but not very
intuitive and rather cumbersome when implemented
in software codes. That is why some authors have
developed equations directly relating position and
incident angles with a particular motion. However,
they were given with specific constraints. For example,
the classic book by Duffie and Beckman18 provides
useful equations but only for surfaces that rotate on
axes parallel to the surfaces. Hence, despite these
antecedents, there is a need to develop a set of
equations for arbitrarily oriented surfaces, also able to
take shading into account, and still easy to implement
in software codes.
Figure 1 describes the solar coordinates referred to a
system whose X, Y and Z axis are pointing,
respectively, to West, South and Zenith. It is not
difficult to deduce these coordinates in relation to solar
elevation, gS, and solar azimuth, cS, angles by
x ¼ cos gSsincS;
y ¼ cos gS coscS; z ¼ singS
(1)
At any given moment, the solar elevation and
azimuth angles are calculated from the equations
sin gS ¼ sin d sinfþ cos d cosf cosv (2)
and
coscS ¼ ðsin gs sinf� sin dÞ=ðcos gs cosfÞ (3)
s
s
z (Zenith)
x (West)
y (South)
Figure 1. Sun coordinates referring to the site
Copyright # 2008 John Wiley & Sons, Ltd.
where d, f and v are, respectively, the latitude, the
declination and the hour angles.
Single vertical axis
For a plane with a fixed slope, bGEN, rotated around a
vertical axis, the angle of incidence is minimized when
the surface azimuth and solar azimuth angles are equal.
Axis rotating angle, vID (‘ID’ means ideal value) and
incidence angle, uS, are easily determined from
Figure 2.
vID ¼ cS (4)
uS ¼ gS � bGEN (5)
Single horizontal axis
For a plane rotated around a horizontal axis and forming
a fixed angle with this axis, bGEN, the angle of incidence
is minimized when the axis, the line to the Sun and the
normal to the surface are on the same plane. Relevant
equations are determined from Figure 3.
tanvID ¼ x
z¼ cos gS sincS
singS
(6)
uS ¼ arc tany
ðx2 þ z2Þ1=2
" #� bGEN (7)
Single tilted axis
For a plane rotated around a single axis oriented to the
south and tilted an angle, bAXIS, and forming a fixed
angle with this axis, bGEN, the angle of incidence is
minimized, again, when the axis, the line to the Sun
and the normal to the surface are on the same plane. In
Figure 2. Vertical tracking: (a) plane of the line to the Sun
and the normal to the surface. (b) View of this plane showing
incidence angle
Prog. Photovolt: Res. Appl. 2008; 16:703–714
DOI: 10.1002/pip
Figure 3. Horizontal tracking: (a) plane of the line to the Sun
and the normal to the surface. (b) View of this plane showing
incidence angle
TRACKING AND GROUND COVER RATIO 705
order to deduce the relevant expressions, it is best to
refer the position of the Sun to the OX0Y0Z0 systemresulting from turning the previous OXYZ system just
an angle bAXIS around the X axis, as shown in Figure 4.
Then, the Sun coordinates at the new system are
given by
x0 ¼ x; y0 ¼ r cosðu þ bAXISÞ;
z0 ¼ r sinðu þ bAXISÞ(8)
where
r ¼ ðy2 þ z2Þ1=2 y u ¼ arc tanðz=yÞ (9)
Now, relevant expressions can be deduced through a
simple analogy with the previous case. Thus,
vID ¼ arc tanx0
z0
� �(10)
uS ¼ arc tany0
ðx02 þ z02Þ1=2
" #� bAXIS (11)
axis
axis
x = x’
y
y’
zz’
Figure 4. Meridian view of Y,Z and Y0,Z0axis
Copyright # 2008 John Wiley & Sons, Ltd.
Two axes
Two-axes tracking is obtained when adding the
capability to rotate the receiving plane around a
second axis normal to the first one to a single axis
tracking. Then, this plane can always be kept normal to
the sun. The new angle, bGEN-TRAC, depends on the
first type of axis. When this is vertical (see Figure 2b)
bGEN�TRAC ¼ gS (12)
Similarly, Figure 3b leads to determining when the
first axis is horizontal
bGEN�TRAC ¼ arc tany
ðx2 þ z2Þ1=2
" #(13)
and when it is tilted
bGEN�TRAC ¼ arc tany0
ðx02 þ z02Þ1=2
" #(14)
SHADING
PV generators often include several trackers. Depend-
ing on the Sun’s position, partial shading between two
adjacent trackers may occur, mainly in the early
morning and late evening.
Single horizontal axis
Figure 5 shows the projection at the XZ plane of two
adjacent trackers, at a distance LEW (distance at the
EW direction), when shading occurs. Horizontal
shadow length, sEW, and shaded fraction, FSEW, are
given by
sEW ¼ 1
cosvID
and
FSEW ¼ max 0; 1� LEW
sEW
� �� � (15)
Single tilted axis
Figure 6 shows a projection at the plane of the Y axis
and normal to the surface when shade in the NS
direction takes place. This figure applies to receptors
rotated around a horizontal axis and forming a fixed
angle, bGEN, with this axis, and also for receptors
rotated around a tilted axis, bAXIS, and coplanar with
Prog. Photovolt: Res. Appl. 2008; 16:703–714
DOI: 10.1002/pip
.
LEW
.1
id
SEW
iz
x
FSE
Figure 5. Two horizontal trackers as projected at the XZ
plane when shading occurs
1
b
β·
706 L. NARVARTE AND E. LORENZO
this axis. In other words, this figure applies when only a
tilt angle, bGEN or bAXIS, differs from zero. Tracker
length and distance, both in the NS direction, are
respectively b and LNS. Then, horizontal shade length,
sNS, and shaded fraction, FSNS, are given by
sNS ¼ b cos bþ sin b
tanB
� �(20)
tanB ¼ ðx2 þ z2Þ1=2
y(21)
FSNS ¼ max 0; 1� b � LNSsNS
� �� �(22)
In these equations b can represent, both, bGEN or
bAXIS.
It can be observed that the following value of LNS
LNS ¼ b cos bþ sin b � tanðfþ dMÞð Þ (23)
just avoids shade at noon for all the days throughout the
year, where dM is the maximum declination angle
throughout the year (dM¼ 23.58).
LNS
b
sNS
B
y
b.FSP
GEN or AXIS
Figure 6. Two horizontal (or tilted) one-axis trackers as
projected at the plane of the Yaxis and normal to the surface,
when shading in the NS direction occurs
Copyright # 2008 John Wiley & Sons, Ltd.
Single vertical axis
Let us consider a set of vertical trackers arranged as
shown in Figure 7.
Figure 8 shows a case of shading in the East–West
direction. The horizontal projection of two adjacent
tracked surfaces and their corresponding mutual
shading are presented in Figure 8a, while the meridian
projection is presented in Figure 8b. Note that
incidences of shading require two simultaneous
conditions. First, the shadowmust point to the adjacent
tracker and, second, its length must be large enough to
reach it. Shadow length and shaded fraction are given,
respectively, by
sEW ¼ s1 þ s2 ¼ b cos bGEN þ b sin bGENcotgS (24)
and
FSEW ¼ FS1 � FS2 (25)
where
FS1 ¼ max 0;s� LEW � sincS
s
� �and
FS2 ¼ max 0;1� LEW � coscS
1
� � (26)
Figure 9 shows a case of shade in the North–South
direction. Strictly speaking, effective shading requires
lEW
lNSN
Figure 7. Parameters of a field with single vertical axis
trackers: tilt angle bGEN; aspect relation b; spacing between
adjacent trackers in North–South and East–West directions
lNS and lEW
Prog. Photovolt: Res. Appl. 2008; 16:703–714
DOI: 10.1002/pip
Figure 8. East–West shadow in ideal single vertical axis
tracking. Projections at (a) Horizontal plane (b) Vertical
plane of the line to the Sun
Figure 9. North–South shade in vertical single axis tracking.
Projections at (a) Horizontal plane (b) Vertical plane of the
line to the Sun
TRACKING AND GROUND COVER RATIO 707
two different conditions: First, the shade borderline
must reach the front of the rear trackers and, second,
shade must impinge on a tracker and not only on the
empty space between two trackers. However, a
reasonable approximation, in particular for relatively
low LEW values (let us say, LEW< 3) consists of only
considering the first condition. Then,
FSNS ¼ max 0; 1� s0
sNS
� �� �(34)
where
s0 ¼ LNS � sincS
sinðp=2� cSÞ(35)
Two axes
Shade considerations already taken into account for
vertical axis trackers can also be applied for two axes
trackers, when considering that the tilt angle is now the
complementary to the Sun’s elevation angle.
Copyright # 2008 John Wiley & Sons, Ltd.
ENERGY YIELD CALCULATION
There are a number of software tools available to
simulate the energy yield of PV grid-connected
systems.19–23 However, none of them can deal with
all of the possible tracking strategies. Hence, we have
developed our own code, comprising the following steps:
1. O
btaining solar radiation site information (the 12monthly averages of the daily horizontal global
irradiation, the daily maximum ambient tempera-
ture and the daily minimum ambient temperature)
from available databases.
2. P
reparation of horizontal (both global and diffusecomponent) irradiance sequences (it can be
selected hourly or minute time intervals).
3. D
etermination of the position of the Sun and thereceiving surface, and the incidence angle.
4. D
etermination of shade.5. T
ransposition from horizontal to inclined irradi-ance values.
6. C
alculation of the losses of dirty PV modules.Prog. Photovolt: Res. Appl. 2008; 16:703–714
DOI: 10.1002/pip
Table I. Models used to simulate the energy yield of a grid-connected PV plant
Step Comment/Reference
1 Solar irradiation site information PVGIS database (for solar radiation)24 and NASA database
(for ambient temperature)25
2 Irradiation to irradiance (horizontal) First, solar radiation components are calculated by means of the Global—
diffuse (Kt, Kd) correlation described by Erbs.26 Second, irradiance
values are derived from those of the daily irradiation, as described by
Collares-Pereira and Ralb.27
3 and 4 Geometry As described in the previous section
5 In-plane irradiance Anisotropic properties of the diffuse solar radiation are considered as
described by Hay and McKay.28
6 Degree of cleanliness and incidence
angle effects
Incidence angle effects are considered in relation with the degree of
cleanliness, as described by Martin and Ruiz.29
7 Shading effects It is considered that a shadow only affects the direct and circumsolar
(a part of the diffuse) components of the solar irradiance.
8 PV array output power PV array efficiency is considered to be linearly related to solar cell temperature.
The constant coefficient depends on the solar cell technology. For most
Si-x modules, the value of �0�5% per 8C is appropriate.
9 Inverter output power Inverter efficiency is considered to vary with load ratio as described by Schmid
and Schmidt.30 This model involves the determination of three dimensionless
parameters characterising the full behaviour of the inverter. These parameters
are here previously determined from the efficiency information provided by
inverter manufactures for several load ratios: typically
5,10,20, 50 and 100%. In this work we have used Ki0¼ 0�004461;Ki1¼ 0�01473;Ki2¼ 0�02093 which correspond to an inverter with a European efficiency¼ 96%.
10 Transformer efficiency versus output power Transformer losses are the sum of the so called stand-by losses and wiring losses
The latter is related to the square output power. Transformer manufactures
provide characteristic information.
708 L. NARVARTE AND E. LORENZO
7. C
Copy
alculation of the losses of PV modules in the
shade.
8. C
alculation of the PV array output power.9. C
alculation of the inverter output power.10. C
alculation of the low voltage/medium voltagetransformer output power.
Table I sets out brief explanations for each step.
It is worth mentioning that soiling impact can be
very significant when dry climates are concerned, as it
is the case in most of Spain. Our code paid particular
attention to this fact, supported by Reference 29 at step
6. This not only allows soiling impact on PV modules
cover transparency to be into consideration but also
soiling impact in angular reflection losses.
SELECTED CASES
Although there are a large variety of trackers, the
tracking strategies considered in this paper are the most
extended ones on the current market: two axes, one
vertical axis, one north–south horizontal axis and one
north–south horizontal axis tilted 208. We have also
right # 2008 John Wiley & Sons, Ltd. Prog. Photovolt: Res. Appl. 2008; 16:703–714
DOI: 10.1002/pip
.
considered, as a reference, the case of a static surface
with an optimal tilt, that is the tilt that maximizes the
annual incident irradiation.
The output power of a PVarray can be described by
P ¼ P� G
G� ð1� bðTc � T�c ÞÞ (36)
where Tc¼ TaþKG, P� is the nominal power of PV
generator, G�¼ 1000W/m2, G is the irradiance on the
PV generator’s surface, K and b are characteristic
parameters of the solar cell technology, and Ta is the
ambient temperature. The good behaviour of this rather
simple approximation to the power output of PV
modules has been validated by recent studies.31
Here we assume that shading affects the direct and
circumsolar components of the incident irradiance, and
not the isotropic components. Hence,
GBS ¼ Bþ Diso þ Dcircumsolar þ R
GAS ¼ ðBþ DcircumsolarÞð1� FSÞ þ Diso þ R
whereGBS andGAS represent the irradiance before and
after shading consideration respectively, and B, D and
R respectively, are the direct, diffuse and reflected
components of the irradiance.
File 1
File 2
+ -Vertical axis
TRACKING AND GROUND COVER RATIO 709
For a given shaded fraction FS, defined as the ratio
between the shaded area and the total area of a PV
generator surface, the power reduction SI, depends on
the particular electric PV system layout (mainly solar
cells series and parallel association, bypass-diode
disposition and inverter arrangement). Whatever be the
case, it encompasses two extremes:
File 3
Horizontal axis
(a) T
Copy
he most optimistic is when the power reduction is
just equal to shaded fraction (SI¼FS).
(b) T
hemost pessimistic is when the mere existence ofany shade fully avoids power (FS> 0! SI¼ 1).
Figure 10. Electrical layout of a PV generator considered
here
This paper analyses these two extremes and the casedetailed in Figure 10, which can be often found in real
PV plants: the PV generator carried by a tracker is
made by the parallel association of three rows, each
made up of the series association of a certain number of
PVmodules. Thus, each time one row gets shaded output,
power reduces by one third. (0<FS< 1/3! SI¼ 1/3;
1/3�FS< 2/3! SI¼ 2/3; 2/3�FS� 1! SI¼ 1).
Land occupation is established by the so called
Ground Cover Ratio, GCR, defined as the ratio
between PV array area to total ground area. For
simplicity, we have restricted our analysis to the
following cases:
– s
ingle vertical axis (with b¼ location latitude) andtwo-axes individual trackers are all square (b¼ 1)
Figure 11. Evolution of yearly energy yield in Almerı
the optimistic shading case and assuming
right # 2008 John Wiley & Sons, Ltd.
and equally spaced over the ground (LNS¼ LEW¼ L;
GCR¼ 1/L2)
– s
ingle pure horizontal axis is understood as infinitelylarge (GCR¼ 1/LEW)
– s
ingle tilted axis is three times longer than wide(b¼ 3) and LNS is selected to avoid shading in this
direction fully, which is close and avoids shading at
the noon winter solstice. It can be shown that
this condition leads to LNS¼ b(cos bþ sin
b � tan(þ dM)), where F is the location latitude
and dM the maximum value of the declination angle
(dM¼ 23�58) (GCR¼ b/(LNS � LEW)).
a for the tracking strategies considered here, for
a constant dirtiness degree of 3%
Prog. Photovolt: Res. Appl. 2008; 16:703–714
DOI: 10.1002/pip
Table II. Yearly energy yield for the tracking strategies considered here and the optimistic shading case for some GCR
values and referenced to the case of an optimally tilted static surface
1/GCR Fixed tilt N–S N–S (08 208) N–S (208 08) Azimuthal Two-axes
1 1 1�037 0�662 0�6662 1 1�177 1�142 1�179 1�048 1�0613 1 1�233 1�270 1�292 1�219 1�2424 1 1�253 1�317 1�332 1�283 1�3225 1 1�263 1�338 1�351 1�320 1�3746 1 1�268 1�351 1�361 1�338 1�4037 1 1�271 1�357 1�367 1�349 1�4248 1 1�274 1�362 1�371 1�356 1�4379 1 1�275 1�365 1�373 1�361 1�44810 1 1�277 1�368 1�376 1�365 1�455
710 L. NARVARTE AND E. LORENZO
Finally, it must be noted that when shading occurs, it
is possible to avoid it by moving the surface angles
away from their ideal values, just enough to get the
shadow borderline to pass over the border of the
adjacent tracker. This is called ‘back-tracking’, and is
in fact implemented in some commercial products.
However, ‘back-tracking’ will be dealt with in a future
paper.
RESULTS
As a representative example, Figure 11 shows the
evolution of the yearly energy yield for the tracking
Figure 12. Evolution of yearly energy yield in Alm
the realistic shading case and assumin
Copyright # 2008 John Wiley & Sons, Ltd.
strategies considered here, as calculated for Almerıa,
for the optimistic case and assuming a constant degree
of dirtiness of 3% (Normal transmittance¼ 0�97).Table II details the numerical results for some GCR
values, referring to the case of an optimally tilted static
surface. Outstanding facts are:
– T
erı
g
here is like a ‘surrounding curve’ describing the
relationship between maximum energy yield and
GCR.
– T
his contour curve roughly, coincides with horizon-tal tracking for 1< 1/GCR< 2, with a north–south
tracking axis tilted 208 for 2< 1/GCR< 4 and with
two-axes tracking for 1/GCR> 4.
a for the tracking strategies considered here, for
a constant dirtiness degree of 3%
Prog. Photovolt: Res. Appl. 2008; 16:703–714
DOI: 10.1002/pip
Table III. Yearly energy yield for the tracking strategies considered here together with the realistic shading case for some
GCR values and referenced to the case of an optimally tilted static surface
1/GCR Estatico N–S N–S (08 208) N–S (208 08) Azimuthal Two-axes
1 1 0�917 0�486 0�4692 1 1�109 1�015 1�062 0�905 0�8873 1 1�195 1�199 1�230 1�103 1�0934 1 1�227 1�271 1�292 1�201 1�1995 1 1�243 1�306 1�321 1�253 1�2696 1 1�252 1�324 1�337 1�283 1�3117 1 1�259 1�336 1�348 1�305 1�3478 1 1�263 1�345 1�355 1�320 1�3719 1 1�266 1�350 1�359 1�332 1�39110 1 1�269 1�354 1�363 1�339 1�405
TRACKING AND GROUND COVER RATIO 711
– W
Co
hen comparing them with static surfaces (1/
GCR¼ 2), energy gains associated to practical
trackers range from 18–25% (horizontal axis, 1/
GCR¼ 2–4) to 37–45% (two axes, 1/GCR¼ 5–10).
– S
ingle vertical axis provides 96% of the energyprovided by two-axes tracking (1/GCR¼ 5).
– T
ilting the PV modules over the horizontal axisleads to very similar results as tilting the axis over
the horizontal.
– N
orth–south axis tilted 208 and single vertical axisshow very similar results when 1/GCR!1.
– N
orth–south axis tilted 208 provides the sameenergy yield as two-axes tracking when 1/GCR¼ 4,
4, and it represents 90% of the maximum energy
Figure 13. Evolution of yearly energy yield in Al
for the pessimistic shading case and assu
pyright # 2008 John Wiley & Sons, Ltd.
yield of a practical tracker (two-axes with 1/
GCR¼ 10).
Figures 12 and Table III, and Figure 13 and Table IV
detail the results of a similar exercise but for the
previously called realistic and pessimistic cases,
respectively. In order to make the reader’s task easier,
Figure 14 puts together the three cases corresponding
to two-axes tracking. It can be seen that the same
comments for the optimistic case apply here with some
additional remarks:
– T
me
mi
he contour curve is made up of the same tracking
strategies, but the more pessimistic the shading case,
rıa for the tracking strategies considered here,
ng a constant dirtiness degree of 3%
Prog. Photovolt: Res. Appl. 2008; 16:703–714
DOI: 10.1002/pip
Table IV. Yearly energy yield for the tracking strategies considered here and the pessimistic shading case for some GCR
values and referenced to the case of an optimally tilted static surface
1/GCR Estatico N–S N–S (08 208) N–S (208 08) Azimuthal Two-axes
1 1 0�576 0�436 0�4292 1 0�952 0�713 0�784 0�692 0�6843 1 1�109 1�034 1�088 0�968 0�9504 1 1�170 1�167 1�202 1�101 1�0835 1 1�201 1�230 1�257 1�176 1�1736 1 1�219 1�267 1�287 1�221 1�2267 1 1�231 1�290 1�307 1�254 1�2718 1 1�240 1�306 1�320 1�276 1�3049 1 1�246 1�318 1�329 1�296 1�33210 1 1�251 1�326 1�336 1�310 1�351
Co
712 L. NARVARTE AND E. LORENZO
the wider the portion of the contour occupied by the
north–south tracking axis tilted 208.
– E nergy gains referenced to static surfaces (1/GCR¼ 2) decrease, ranging from 11% (horizontal
axis, 1/GCR¼ 2) to 40% (two axes, 1/GCR¼ 10)
for the realistic case, and from �5 to 35% for the
pessimistic case.
– W
hen comparing with the optimistic shading case(1/GCR¼ 5) for the two-axes tracking, the energy
yield of the realistic case is the 92% and the pessi-
mistic case is the 85%.
Finally, Table V sets out the results of extending the
simulation exercise for the realistic case to several
Figure 14. Comparison of yearly energy yield in A
cases consid
pyright # 2008 John Wiley & Sons, Ltd.
Mediterranean places. It can be noted that, despite
relatively large variations in latitude and clearness
index, the previous comments essentially remain the
same. It is worth to note that these results can be
different than the ones given by other available tools. It
is not easy to determine the causes of these differences
due to the details of the internal models of these tools
are out of the knowledge of the authors of this paper.
For example, the impact of shadows in the output
power of PV modules is highly dependant of the
shading case, as it has been shown. The model to
characterize the degree of cleanliness of PV modules
and the incidence angle effects can also explain some
of these differences.
lmerıa for the two-axes tracking, for the shading
ered here
Prog. Photovolt: Res. Appl. 2008; 16:703–714
DOI: 10.1002/pip
Table V. Yearly energy yield for the tracking strategies considered here and the realistic shading case for some GCR values
and referenced to the case of an optimally tilted static surface. Comparison of some Mediterranean locations
Location Lat (8) Kt N–S N–S (08 208) N–S (208 08) Azimuthal Two-axes
1/GCR 2 5 10 2 5 10 2 5 10 2 5 10 2 5 10
Sevilla 37�5 0�587 1�12 1�25 1�28 1�02 1�31 1�36 1�06 1�33 1�37 0�91 1�26 1�35 0�89 1�27 1�41Toledo 39�8 0�571 1�11 1�25 1�28 0�99 1�31 1�36 1�04 1�32 1�37 0�90 1�26 1�36 0�89 1�27 1�41Tel-Aviv 32�1 0�552 1�16 1�28 1�31 1�09 1�32 1�36 1�12 1�34 1�37 0�96 1�26 1�32 0�94 1�28 1�40Crete 35�3 0�546 1�15 1�28 1�30 1�06 1�32 1�36 1�09 1�33 1�36 0�96 1�27 1�34 0�94 1�28 1�40Alger 36�7 0�537 1�13 1�25 1�28 1�03 1�31 1�35 1�07 1�32 1�36 0�92 1�26 1�34 0�90 1�27 1�40Larisa 39�6 0�498 1�14 1�26 1�27 1�02 1�31 1�35 1�06 1�31 1�35 0�95 1�27 1�35 0�93 1�28 1�40Toulouse 43�6 0�492 1�10 1�22 1�24 0�94 1�28 1�34 0�99 1�29 1�33 0�89 1�25 1�35 0�87 1�25 1�39
TRACKING AND GROUND COVER RATIO 713
CONCLUSIONS
A comprehensive set of equations describing the Sun’s
tracking geometry and a complete procedure to
calculate the energy yield of a flat PV module grid-
connected system have been presented. Based on
these, the relationship between the yearly energy gains
and land occupation has been analysed for several
tracking strategies. The outstanding results are as
follows:
– T
Co
here is a ‘surrounding curve’ describing the
relationship between maximum energy yield and
GCR, made up of three different tracking strategies;
horizontal tracking for 1< 1/GCR< 2, north–south
axis tilted 208 tracking for 2< 1/GCR< 4 and two-
axes tracking for 1/GCR> 4. When more pessi-
mistic shading models are used, the contour curve
remains but the portion of the contour occupied by
the north–south axis tilted 208 tracking is wider.
– T
he energy gains associated to practical trackersreferenced to static surfaces (1/GCR¼ 2) has been
calculated. For the optimistic shading case, these
gains range from 18–25% (horizontal axis, 1/
GCR¼ 2–4) to 37–45% (two axes, 1/GCR¼ 5–10).
– T
ilting the PV modules over the horizontal axisleads to very similar results as tilting the axis over
the horizontal.
– T
he influence of the shading model is more import-ant when the land occupation in larger. As a practical
example (1/GCR¼ 5), and taking the optimistic
shading case as a reference for the two-axes track-
ing, the energy yield of the realistic case is 92% and
the pessimistic case is 85%.
The extension of this simulation exercise to several
European places has shown that despite the relatively
pyright # 2008 John Wiley & Sons, Ltd.
large variations in latitude and clearness index, the
previous results are applicable.
REFERENCES
1. Wenger HJ, Jennings C, Iannucci JJ. Carrisa Plains PV
power plant performance, Conference Record of the
Twenty First IEEE, Photovoltaic Specialists Conference,
1990.
2. Lorenzo E, Maquedano C, Valera P. Operacional results
of the 100 kWp tracking PV plant at Toledo-PV project.
Proceedings of the 13th European PV Solar Energy
Conference, Nice, 1995; 2433–2436.
3. Jennings C, Farmer B, Townsend T, Hutchinson P,
Gough J, Shipman D. PVUSA—The first decade of
experience. Proceedings of the 25th IEEE PVSC,
Washington, 1996; 1513–1516.
4. Alonso-Abella M, Chenlo F, Vela N, Chamberlain J,
Arroyo R, Alonso F. Toledo PV plant 1 MWp—10 years
of operation. Proceedings of the 20th European PV Solar
Energy Conference, Barcelona, 2005; 2454–2457.
5. Illiceto A, Previ A. Progress report on ENEL’s 3�3 MWp
PV plant. Proceedings of the 12th PV Solar Energy
Conference, Amsterdam, 1994; 1167–1170.
6. Stern M, West R. Development of a low cost integrated
15kWAC solar tracking sub-array for grid-connected PV
power system applications. Proceedings of the 25th
IEEE PVSC, Washington, 1996; 1369–1372.
7. Ito M, Kato K, Komoto K, Kichimi T, Sugihara H,
Kurokawa K. Comparative study of fixed and tracking
system of very large-scale PV (VLS-PV) systems in the
World deserts. Proceedings of the 19th European PV
Solar Energy Conference, Paris, 2004; 2113–2116.
8. Klotz F, Schroeder S,MohringH. Light weight PV-louvres
for building integration. PV-Light EU-Project (FP5
NN�2001-550). Proceedings of the 19th European PV
Solar Energy Conference, Paris, 2004; 3154–3157.
9. Kimber A,Mitchell L. Performance evaluation standards
for large grid-connected PV systems in the United States
Prog. Photovolt: Res. Appl. 2008; 16:703–714
DOI: 10.1002/pip
714 L. NARVARTE AND E. LORENZO
and Germany. Proceedings of the 20th European PV
Solar Energy Conference, Barcelona, 2005; 3229–3232.
10. Lorenzo E, Perez M, Ezpeleta A, Acedo J. Design of
tracking photovoltaic systems with a single vertical axis.
Progress in Photovoltaics: Research and Applications
2002; 10: 533–543.
11. Guelbenzu E, Cueli A, Lagunas A. 1,2 MW photovoltaic
plant with sun tracking on all generation systems. Pro-
ceedings of the 19th European PV Solar Energy Con-
ference, Paris, 2004; 2258–2261.
12. Zdanowicz T, Rodziewicz T, Zabkowska M. Effective-
ness of the sun tracking option during different insola-
tion periods. Proceedings of the 19th European PV Solar
Energy Conference, Paris, 2004; 2580–2583.
13. Sureda F, Salas J, Gil J, Niesing H, Cantos S, Vidal J.
FSA SystemTM: an autonomous small scale power sys-
tem for distributed generation based on PV with an
effective demand side management. Proceedings of
the 20th European PV Solar Energy Conference, Bar-
celona, 2005; 2228–2230.
14. Comsat M, Calin R, Visa I, Duta A, Meester B. A
tracking system design for PV panels. Proceedings of
the 20th European PV Solar Energy Conference, Bar-
celona, 2005; 2212–2215.
15. Luque-Heredia I, Moreno J, Quemere G, Cervantes R,
Magalhaes P. SunDogTM STCU: a generic sun tracking
control unit for concentration technologies. Proceedings
of the 20th European PV Solar Energy Conference,
Barcelona, 2005; 2047–2050.
16. Siemer J. Market survey on tracking systems. A market
offering many choices. Photon International October
2007; 114–140.
17. Collins GW II. The Foundations of Celestial Mechanics.
Case Western Reserve University, 2004. URL http://
ads.harvard.edu/books/1989fcm.book/
18. Duffie JA, Beckman WA. Solar Engineering of Thermal
Processes. John Wiley & Sons, Inc.: New York, 1991.
19. PV Design Pro, Maui Solar Energy Software Corpor-
ation. Haiku. USA.
20. Menicucci D, Fernandez J. User’s manual for
PVFORM: a photovoltaic system simulation program
for stand-alone and grid-interactive applications, SA-
Copyright # 2008 John Wiley & Sons, Ltd.
ND85-0376. Sandia National Laboratories, Albuquer-
que, 1988.
21. Wenger H, Gordon J. PVGRID: a user-friendly PC
software package for predicting utility-intertie photo-
voltaic system performance, ISES SolarWorld Congress,
Denver, 1991.
22. Marion B, Anderberg M. PVWATTS—An online per-
formance calculator for grid-connected PV systems.
Proceedings of the ASES Solar 2000 Conference, Madi-
son, 2000; 119–121.
23. Eicker U, Pietruschka D, Schumacher J, Fernandes J,
Feldmann T, Bollin E. Improving the energy yield of PV
power plants through internet based simulation, monitor-
ing and visualisation. Proceedings of the 20th European
PV Solar Energy Conference, Barcelona, 2005; 2670–
2673.
24. Available at http://re.jrc.ec.europa.eu/pvgis/apps/rad-
month.php
25. Available at http://eosweb.larc.nasa.gov/sse
26. Erbs et al. Estimation of the diffuse radiation fraction for
hourly, daily and monthly-average global radiation.
Solar Energy 1982; 28: 293–302.
27. Collares-Pereira M, Ralb A. The average distribution of
solar radiation—Correlation between diffuse and hemi-
spherical and between daily and hourly insolation
values. Solar Energy 1979; 22: 155–164.
28. Hay JE, McKay DC. Estimating solar irradiance on
inclined surfaces: a review and assessment of method-
ologies. International Journal of Solar Energy 1985; 3:
203–240.
29. Martin N, Ruiz J. Calculation of the PV modules angular
losses Ander field conditions by jeans o fan analytical
model. Solar EnergyMaterials and Solar Cells 2001; 70:
25–2538.
30. Schmid J, Schmidt H. Inverters for photovoltaic systems,
5th Contractor’s Meeting of the PV Demostration Pro-
jects. Ispra, 1991; 122–132.
31. Fuentes M, Nofuentes G, Aguilera J, Talavera DL,
Castro M. Application and validation of algebraic
methods to predict the behaviour of crystalline silicon
PV modules in Mediterranean climates. Solar Energy
2007; 81: 1396–1408.
Prog. Photovolt: Res. Appl. 2008; 16:703–714
DOI: 10.1002/pip