2004 1st International Conference on Electrical and Electronics Engineering

Optimization of the Solar Energy Collection

in Tracking and Non-Tracking Photovoltaic Solar System

Pave1 Yu.Vorobiev Jesus Gonzalez-Hemhdez ,Yuri V. Vorobiev

Facultad de Ingenieria Unidad Queretao of CINVESTAV
Universidad Aut6noma de Queretaro CINVESTAV-IPN
C.P.76010, Queretaro, Qro., MEXICO Queretaro-76230, QRO.,MEXICO

Abstract - Experimental and theoretical analysis is made of To explain the difference between the AM0 value
the conditions of collection of solar radiation in solar tracking (cosmic intensity F) and those related to the Earth’s surface,
system without concentration, and in ooo-trackiug system with one has to take into account, except for the losses in passing
standard and bifacial photovoltaic solar panel. For the through atmosphere, also reflection (back scattering, more
experiments with solar tracking, an original Sun tracker was exactly) from it; this effect in rough approximation could be
developed. The tracking effect calculated and measured shows
an increase in energy collection around 30 - 40 %a; bifacial
considered as independent on angle 4, Thus, we obtain an
panels with a reflector collecting solar radiation for the rear approximate dependence of the Sun radiation intensity at the
face give the corresponding increase of 50 - 60 % for the same Earth’s surface I, upon the angle 4 as follows:
panel area.

1. INTRODUCTION
where the constant C takes account of back scattering, and B
The efficiency of solar energy convertion by
is proportional to the atmosphere thickness. It i s evident that
photovoltaic (PV) panels is the problem which constantly is
(1) is an approximate expression valid only for relatively
in a focus of attention of scientists and technologists in the
smaIl Iosses in atmosphere; the better approximation is given
field. One of its many aspects is the concentration of solar
by the expression
radiation andlor tracking of the Sun, to increase the total
energy collected by a panel for conversion. Significant
progress achieved here during the last decade was mainly b = CF exp (-ad)
due to employment of semiconductor high technology and
high degree of concentration. The purpose of this paper is to where ais the effective coefficient of absorption +
analyse the possibilities of increase the efficiency o f scattering in the atmosphere, and d - its effective thickness
utilization of solar energy in more practical way, using for a given angle (the term “effective” here means that we
traditional and bifacial PV panels, low concentration substitute the real atmosphere with height-dependent
systems and economic Sun trackers. parameters by the uniform one having the same total
absorption). For a d << 1, we put
TI. MODELLING OF DAILY VARIATIONS OF SOLAR exp (-ad) = 1 - ad, (2*)
RADIATION ”TENSITY
To be able to evaluate correctly the amount of solar so that ad in (2) will be equivalent to B/cos$ in (1). If we
energy collected by a solar module, one has to know the neglect the Earth’s curvature, the effective atmosphere
dependence of solar radiation intensity (i.e. irradiance) for a thickness d could be written as h/cos+, where h is this
given module position upon the angle 4 between the Sun thickness for zero angle. It means that B = ah, and (2) could
flux direction and zenith. There are many publications on the be rewritten as
subject, including different models (for example, [ 1-61)
which usually demand very detailed information about the IO = AF exp (-BkosQ) (3)
state of atmosphere to be used (like light dispersion and
absorption caused by water vapor, ozone layer, aerosol, l%e constants A and B in (1) and (3) we find, averaging the
etc.); that makes their utilization for quick estimations data found from the pairs of algebraic equations obtained by
practically impossible. insertion in (1) or (3) the data for AM1 and AM1.5, and then
We present here a simple model to calculate the for AM1 and AM2. In case of equation (1) this procedure
dependence mentioned which is based upon the data known gives
for the solar radiation intensity at different air mass: AM0
(1367W/m2), AM1 (925 Wlm’), AM1.5 (844 W/m2) and I,, = I 140(1 - O.I88/cos$) W/m2 (I*)
AM2 (691 W/m2); it is generally accepted that the air mass
index for the last three values corresponds to l/cos+, the Using for calculations the equation (3), we get slightly
latter expression is easily associated with an increase of the different values of constants:
path of the solar flux in atmosphere with increase of Q if one
neglect the Earth’s curvature (in other words, it is a good b = 1200 exp (-O.~~/COSL$) W/m2 (3*)
approximation for relatively small values of Q; this is one of
the points to be analyzed). The Fig. 1 shows the angular dependencies
corresponding to (1*) and (3*) as c w e s 1 (squares) and 2

0-7803-8531-41041$20.00 02004 IEEE 310

(circles); it is evident that for angle not exceeding 60°, the
curves practically coinside; for larger angles the curve 1 d = R cos$ [( 1 + 2h/R COS^$)''^ - 11. (5)
goes down more rapidly, reaching zero value at angle
smaller than 90’ which is clearly incorrect. Thus, expression An analysis of the expression obtained shows that for
(3) is the better approximation here, meaning that the small angles Q (cos4 is close to 1, and since 2h << R,we can
absorption in atmosphere at large angles is not small and the take that (1 f 2hlR cos2Q)1’2= 1 + MZ cosz$) the value of d
approximation (2*) is not a good one. We can also note that is approximately equal to h/cos$, in a good agreement with
the curves 1 and 2 in Fig. 1 give for the AMO, AM1.5 and the discussion above. For angle values close to 90°, (5) gives
AM2 the values around 925, 840 and 700 W/m2 the finite value (2Rh)’”, contrary to the infinity given by
correspondingly, which we consider as good enough h/cos$.
approximation. Having introduced the value of d given by (5) in the
One can see that the curve 2 in Fig. 1, although gives the above expression ( 2 ) and using, as before, the notation B =
larger values at large angles than the curve 1, is also going ah,we obtain, instead of [3*), the new expression
down to zero at angles smaller than 90”, which clearly could
be attributed to the effect of the Earth’s curvature: W h 2 (6)
[( 1+2hiRco~*~)”*-l]lh)
Io=1200exp(-0.26Rc~~~
neglecting it, we get the larger atmosphere thickness at large
angles than the real one. This expression couId be considered in our model as the
To take account of this factor, we have to find the more best approximation for the angular dependence of solar
realistic angular dependence of the effective atmospheric radiation. It contains the effective atmospheric thickness h,
thickness. Fig. 2 gives a sketch of the Earth-atmosphere but since h << R, for small angles its value is not essential,
system, the point 0 corresponds to the observer’s position and for large angles it ought to be found from comparison
(solar flux comes from the left, as shown by solid arrows); with the experiment. For initial estimation, we take h = 50
the values of h and d are indicated. Using the theorem km; it gave us the curve 3 (triangles) in Fig. 1. It is seen that
stating that at intersection of a khorde with a diameter, the for angles smaller than 60 this curve practically coinsides
products of their parts are equal, we write with curves 1 and 2 ; for larger angles, curve 3 goes above
the other curves, as it should be.
hw(2R+h)= d*(OL+d), (4) In the following sections we present the calculated and
experimental data for the total energy obtained by the panel
R being the Earth radius. during the day, for two cases: a) immobile panel, orientated
*ml in most profitable manner, and b) the panel with orientation
following (tracking) the Sun, i.e. always normal to the
radiation flux. The data for the last case ought to be
compared with the upper curve in Fig. 1. To be able to
discuss the f i t case, we have drawn the lowest curve in Fig.
1 corresponding to Io cos4 where Iois given by (6).

III. EXPERIMENT: RESULTS AND DISCUSSION

o a a m m i m
*@
A. Two-Axis Solar Tracking and its Effect.
The tracking system designed and made uses the solar
Fig. 1. Angular dependence of direct solar irradiance of a plane surface panel MDSR-65 elaborated by Russian plant OKBKZ,
normal to radiation tlw
Moscow, has two axis of rotation (Fig. 3). The system is
equipped with two geared servo motors (Colman EYQF-
33300-661 and Globe 407A-350); 2 pairs of phototransistors
(1 pair for each axis) are positioned in such a way that the
difference of photoresponce in each pairs is zero when the
e- panel is orientated towards the Sun, and grows with an
increase of angle of desorientation. The system monitoring is
*- based on microcontroller PIC16f877 with 4 input channels
(one for each sensor); the input analog signals in each
channel are digitalized with resolution of 8 bits, the
-= .- differential signal for each pair is calculated, and the
proportional output PWP signal (pulse width modulated)
applied to the motors, until the correct position is reached.
After that, the system enters the “sleeping state” for the time
interval X which could be chosen depending on the accuracy
Fig. 2. Scheme of Earth with atmosphere
of tracking necessary.

The triangle OLM is rectangular, so OL = 2R cos$;

neglecting h in comparison with ZR,we get ffom (4)

31 1
 6 10 12 14 16 18 23
Tim h K

Fig. 3 . Photo of solar tracking system Fig. 4. Experimental and calculated irradiance (SM text).

The squares in Fig. 4 show the transformed theoretical upper

In our case of a stationary system, the value of X = 20
curve of Fig. 1, adjusted in a way that its maximum
min was taken, giving the tracking accuracy (i.e. deviation
coinsides with the maximum of experimental curves, the left
of radiation intensity from maximum) not larger than 0.5 %:
and right parts are symmetrical in relation to this point, and
the largest corresponding Of 4 guring this the angular scale is converted to time $0that 90" correspond
(at s m e r time) is gad and cos 6o 0.995. This to 6.72 h, One could see that theoretical and experimental
tracking system was for Use with the data have a reasonable coincidence, although theoretical
concentrator; the corresponding results will be Published
curve goes a linle lower than the at large angles
separately, here we discuss only the tracking effects without ~

Y.
concentration.
The experimental effect of tracking in our case (i.e. the
To calculate the total solar energy captured daily by the
ratio of areas under the upper and lower curves in Fig. 4) is
Sun-tracking panel, one needs to change the horizontal axis
1.4, a little higher than the calculated one. The total energy
in Fig. 1 from angle 4 to corresponding time t, find the area collected by tracking solar module is 6.5 k W h per day per
under the I(t) curve and double it. For non-tracking panel square meter, for non-tracking module it is 4.65 kwh.
placed horizontally, the irradiation at arbitrary angle (I is For the case of winter insolation at the same latitude
proportional to cos+ (the lowest curve in Fig. 1). The ratio of (winter solstice, illumination from the left in Fig. 2) our
the corresponding areas is 1.354 which means that the analysis gives the daytime o f t * = 12 - 24 tgZ O h = 10.56
calculated tracking effect in the case examined (i.e. that of hrs; the maximum irradiance corresponds to the angle of 47"
the vertical Sun orbite) constitutes 35.4 %. in Fig. 1, and the part of the upper curve between 47 and 90
Experimental conditions which correspond to Fig. 1 (i.e. grad will describe the time dependence of insolation from
with the maximum at 4 = 0 ) could be realized at summer noon until the end of the day. The corresponding calculated
time, around June 22 (summer solstice), in places of Earth dependence for the radiation flux captured by the tracking
with North latitude close to 8=23.5', like Guadalajara, system is shown by squares (the curve starts at irradiance
Leon or Queretaro in Mexico. Our measurements were made 825 W/mz) in Fig. 5, where At is the time difference fiom
in Queretaro on June 19, 2003, the results are shown in Fig. noon.
4 (triangles with tracking, circles without it). Fig. 2 gives the The lowest curve in Fig. 5 gives the irradiance at winter
scheme of experiment: line A-A denotes the Earth rotation solstice of non-tracking system orientated in the most
axis, Sun flux comes from the right (dashed arrows), the favorable way - normal to the solar fluxat noon, i.e. at 47'
vertical line B-B indicates the boundary between illuminated to the Earth surface. During the day, its angle in relation to
and dark parts of Earth (day-night), 0 3 is the observation this flux varies fiom cp = 0 to 'p = 90°, and multiplication of
point at noon which with time changes its position in
the upper curve by corresponding coscp gives the irradiance.
relation to Sun following the circular line Os - 0 (the end of
a solar day when Q = 90' corresponds to the point 01).
To estimate the duration of a solar day at these
conditions, we have to calculate the distance between the
points 01and 02 of the circle mentioned; a reasonable
approximation could be made on the basis of triangle
0,01C,neglecting the Earth curvature at the corresponding
region. Thus we get 1 = OIOz = Rsine tg9, and the duration
of the day in hours t, = 12 + 24.2 1/2nRcos e = 12 + 24 t$
O/x. In the case considered, the day will be 13.44 hrs.

Fig. 5 . h d i a n c e at different periods (see text)

312
In this case, the calculated tracking effect is 1.28, lower than
in summer case. The total energy produced by tracking solar
module is 3.6 kwh per day per square meter, for non-
tracking module it is 2.8 kWh.
In a similar way, we obtain the irradiance at equinox
time, when the smallest value of 4 at the latitude accepted is
2 3 9 , and the solar day time is 12 hrs. The calculated data
are shown in Fig. 5 by diamonds (the highest curve starting
at irradiance 910 W h 2 ) for tracking module, and by the
curve with crosses starting at the same points for non-
tracking one, orientated at 23.5' to the surface and normal to
solar flux at noon. Two other curves in Fig. 5 (triangles up
and down) are experimental (Queretaro, Mexico, March 20,
2003); they go a little lower than the calculated curves, but
reasonably close to them. The experimental tracking effect is
1.33, and the total energy produced by tracking solar module
is 4.8 kwh per day per square meter, for non-tracking
module it is 3.6 kwh.

B. Non-Tracking System with Sifncial PV PaneL Fig. 7. Two bifacial solar panels with reflectors on the roof of rural school
We see from the discussion above that the tracking effect
in absence of solar concentration is around 30 %, i.e. not To make the reflector cheaper, we substituted cylindric
very high. There exists another possibility to get more power surface shown by solid line with plane stainless steel panels
fiom PV soIar panel of a fixed area - to use bifacial panels as shown by the dashed lines. The actual system with two
[7,8]) sensitive to illumination from both sides; usually, the bifacial panels and reflectors (the other components are not
rear face is able to give 50 - 60 % of the power produced by discussed here) was installed in one of the rural schools in
the upper one. These panel demand some arrangements to the state of Queretaro, Mexico, in Augist 2002 (Fig. 71, and
collect solar radiation for the rear face. One of the since then it is constantly used providing enough energy for
approaches [SI is to use the diffuse reflectance from sufiaces functioning of the receptor of the satellite educational
below the panel, and to paint them correspondingly. This programs and TV-video set 6 hours a day. Our analysis of
method gives good results; however, it needs relatively large the systems performance shows that the panel with
areas of reflecting surfaces, which may be not practical; eEciency of 10 % and area 0.6 m2 having capacity to
besides, the paint lasts less than the PV panel. We employed generate 60 W by front face and 40 W for rear one at AM1.5
the technique demanding smaller area, which in fact is illumination conditions, produces 100 W in total, giving 0.6
concentrating reflector placed below the PV panel. Fig. 6 kWh of electric energy at average summer day, which is
shows a scheme of the reflector, which ideally is a cylindric better than the data given above for solar tracking panel of
or spherical mirror M (the last for a module of the same efficiency.
approximately squared shape), with a panel (PV)placed in a
focal point F, at half a radius from the optical center 0. IV.CONCLUSIONS
This reflector with proportions corresponding to our Experimental and theoretical investigations made lead us
scheme will collect radiation on the rear face of the panel to the conclusion that the use o f solar tracking PV panels in
within an angle 20 determined by the dimensions of the absence of solar concentration gives relatively small increase
panel and the reflector focal distance, 0 gives deviation in (around 30 %) in the solar energy collection and electric
position of Sun fiom zenith. energy production, and therefore could be practical only
with very economic tracking systems. Larger effect could be
achieved with bifacial PV panels, which production cost is
not much higher than that of standard modules of the same
area, and an increase in energy production caused by
effective use of a rear face with a simple system of flat
mirrors could be 50 - 60 %.

V. AKNOWLEDGEMENTS
The authors would like to aknowledge the financial
support by CONACYT of Mexico and CONCYTEQ,
Queretaro.

Fig. 6. Optical scheme of reflector for bifacial solar panel

31 3
 VI. REFERENCES
[I]. N. Robinson, Solar Radiation, Elsevier Publishing,
1971.
[2]. RE. Bird, C. Riordan, “Simple Solar Spectral Model
for Direct and Diffuse Irradiance”, J. Climate Appl.
Meteor., vol. 25, 1986, pp. 87-97.
131. J.A. Duffie, W.A. Beckman, Solar Evgineering of
Thermal Processes. 2“d ed., Wiley-Interscience,
N.Y.; 1991.
[4]. T.M.Klutcher, “Evaluating Model to Predict
Insolation on Tilted Surfaces”, Solar Energy, vol. 23,
1979, pp. 111-123.
IS]. V.V. Satyamurty, P.K.Lahiri, “Estimation of
Symmetric and Asymmetric Hourly Global and
Diffuse Radiation from daily Values”, Solor Energy,
vol. 48, 1992, pp. 7-14.
[6]. Z.Sen, “Fuzzy Algorithm for Estimation of Solar
Irradiation from Sunshine Duration”, Solor
Energy, vol. 68, 1998, pp. 39-49.
[7]. A. Luque, A. Cuevas, J.M. Ruiz, “Double-sided n’-p-
n+ Solar Cell for Bifacial Concentration”, Sol.
Cells, vol. 2, 1980,pp. 151-166.
[XI. A. Luque, E. Lorenzo and G . Sala, “Diffusing
Reflectors for Bifacial Photovoltaic Panels”, ”, Sol.
Cells, vol. 13, 1984-1985, pp. 277-292.

314