Solar Energy 77 (2004) 269–280

www.elsevier.com/locate/solener

Design of a novel passive solar tracker

M.J. Clifford *, D. Eastwood
School of Mechanical, Materials and Manufacturing Engineering, The University of Nottingham, University Park,
Nottingham NG7 2RD, UK

Received 15 August 2003; revised 24 March 2004; accepted for publication 1 June 2004
Available online 28 July 2004

Communicated by: Associate Editor Gabriela Alvarez-Garcia

Abstract

A novel, low cost solar tracker suitable for use in equatorial regions around the world is presented. The solar tracker
is passively activated by aluminium/steel bimetallic strips and controlled by a viscous damper. Computer modelling pre-
dicts an increase in efficiency of up to 23% over fixed solar panels. Experimental testing shows excellent agreement with
the computer model. In addition, further developments to the design are critically evaluated in terms of complexity and
benefit.

2004 Elsevier Ltd. All rights reserved.

Keywords: Tracker; Passive; Appropriate technology

1. Introduction
There are numerous ways to convert the solar energy
into either electricity or heat, be it on an industrial or
commercial scale (Wieder, 1982; Markvart, 1994). One
common method is through the use of photovoltaic
(PV) cells. The power output of PV cells depends on a
number of factors. These include the operating temper-
ature, irradiance and incident angle of the solar radia-
tion. The daily average output of the PV cells can be
enhanced by a solar tracker, which forces sunlight to
be incident normally (perpendicularly) to the PV cell
at all times, mimicking the behaviour of certain flowers
*
Corresponding author. Tel.: +44 115 846 6134; fax: +44
115 951 3800.
E-mail address: mike.clifford@nottingham.ac.uk (M.J.
Clifford).
which follow the sun during the day (Galen and Stanton,
2003). The presence of a solar tracker is not essential for
the operation of a solar panel, but without it, perform-
ance is reduced. Tests have shown that up to 40% extra
power can be produced per annum using a variable ele-
vation solar tracker (Lorenzo et al., 2002; Patil et al.,
1997; Nuwayhid et al., 2001). Some current devices
change the orientation of the PV cell, but this need not
necessarily be the only method (Mccluney, 1983).
An ideal tracker would allow the PV cell to accu-
rately point towards the sun, compensating for both
changes in the altitude angle of the sun (throughout
the day) latitudinal offset of the sun (during seasonal
changes) and changes in azimuth angle. The slow move-
ment of the sun requires a damped system that will also
respond slowly and avoid an oscillatory movement.
Other desirable aspects would include the nocturnal
repositioning of the solar tracker to anticipate the align-
ment of sunrise, opposite to that of the previous dayÕs

0038-092X/$ - see front matter
2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.solener.2004.06.009
270 M.J. Clifford, D. Eastwood / Solar Energy 77 (2004) 269–280

sunset, reducing energy losses in the morning. Various

sensing equipment and actuating devices, both electrical
(active) and mechanical (passive), are in use today
(Helwa et al., 2000a,b).
The current technology on market can be categorised
as follows.

1.1. Electronic (active) trackers

A common solution is that of a centrally pivoted PV

cell being moved, about this pivot, by one or more mo-
tors linked to an electronic sensor (Abouzeid, 2001; Sal-
awu and Oduyemi, 1987; Baz et al., 1984). This is
probably the simplest electronic method available, with
the motor being powered by the cell itself. The panel is
positioned out of reach, to avoid human interference,
although leaving it more susceptible to wind drag. How- Fig. 1. An electronic (active) solar tracker. Control software
ever, as PV power is reduced, this is not a very elegant aims to minimise the angle between the line of the sun and a
solution. Another type of electric solar tracker is shown face perpendicular to the panel using a sensor.
in Fig. 1 and includes a sensor that aims to minimise the
angle between the line of the sun and a face perpendicu-
lar to the panel. When this angle is reduced to zero the
sunlight strikes the panel at 90. Also, in this design,
the central pivot was not horizontal––one bearing was
nearer the ground than the other––giving the correct ele-
vation for non-equatorial locations. Again the panel it-
self produced the power for the motor.
Electronic solutions command the majority of the
market for solar trackers, but they deplete the power
produced by the PV panel for its own operation, and
also have the added expense of fitting and maintaining
an electric motor and control system.

1.2. Mechanical (passive) trackers

A passive solar tracker supplied by Zomeworks

Ltd. 1 is illustrated in Fig. 2. Two identical cylindrical
tubes (each at either side of the panel and equal dis-
tances from the central pivot) are filled with a fluid un-
der partial pressure. Using suitably placed shades, the
sun heats the fluid causing evaporation and transfer
from one cylinder to the other. This mass imbalance is
used to move the solar panel. Damping is used to limit
the speed of movement. This simple system can be made
relatively cheaply and uses none of the PV cellÕs power.
However, it begins each day pointing in the wrong direc-
tion, losing sight of the sun as it attempts to reposition
itself. The use of refrigerants to fill the cylinders needs
careful consideration. Despite these draw backs, it is a
commonly used method.
1
www.zomeworks.com/solar/trackrack/trackrack.html,
accessed 22/11/2002.
Fig. 2. A passive solar tracker using two identical cylindrical
tubes filled with a fluid under partial pressure.
Tests have shown that passive trackers have been
found to be comparable to electrically based systems
in terms of performance (Radajewski, 1987; Mccluney,
1983), but even though they are often less expensive,
they have not yet been widely accepted by the consumer.
1.3. The importance of geographical location on solar
tracker design
As solar irradiance, access to technology, tempera-
ture and precipitation levels differ throughout the world;
it is useful to consider some of the solar tracking
options available with respect to geographical location.
Three cases are considered: Brazzaville, Montreal and
Nottingham.
 M.J. Clifford, D. Eastwood / Solar Energy 77 (2004) 269–280
Brazzaville, Congo-Republic; 4.3 N, 15.2 E. This
developing country is close to the equator and hence
the altitude angle in the middle of the day would consist-
ently be very close to 90. Temperatures in this area
show low variability throughout the year making it eas-
ier to design a tracking system that could be effective all
year round. Limited access to technology favours a sim-
ple mechanical tracker that could be easily repaired with
basic workshop facilities over an electronic solution that
could not be maintained locally.
Montreal, Canada; 45.5 N, 73.6 W. This area experi-
ences a different set of challenges compared to an equa-
torial location. Summer days have long periods of
daylight, although the temperatures and solar intensities
are not as great as those at the equator. These factors re-
quire a tracker to be designed that is able to track
for long periods-maximising summer daylight absorp-
tion. Conversely, on short winter days, the PV cells
would be unable to produce much electricity. Also since
the sun, in both summer and winter, is considerably
lower in the sky, a single axis system would be
inappropriate.
Nottingham, England; 53.0 N, 1.2 W. The challenges
presented by this area are again different. Other factors
such as precipitation would pose more of a maintenance
problem and may lower tracking accuracy. Also, cloud
cover, even in summer months, would leave a passive
tracker unable to locate the sun accurately. In this case
an electronic device may be the best solution.
In this paper, a novel passive solar tracker is designed
for use in developing countries, such as Congo, having
limited access to technology. A design based on the ther-
mal expansion of bimetallic strips is developed in Sec-
tion 2. The concept is evaluated using a computer
model which is developed in Section 3. Design parame-
ters are set in Section 4 using the results of some sensi-
tivity analysis. A prototype tracker is constructed
(Section 5) and tested under artificial lighting conditions
in Section 6. Finally, conclusions and options for further
development of the system are discussed.
2. Design considerations
Although mechanical solar trackers exist, designs
based on partially pressurised fluid are far from ideal.
It may also be difficult to locate a suitable fluid in the
developing world and harder still to pressurise it, inject
it into a container and seal it. An alternative is to design
a system based on the thermal expansion of bimetallic
strips. The physics behind expanding metals indicate
that although generated deflections are small, the corre-
sponding forces are large. A bimetallic strip has obvious
benefits for use in a solar tracker––the strip will bend in
one direction with heat and in the other when cold. Due
to the extremely low temperatures in some African
271
nights, this provides significant scope for a method of
automatic return after the end of the day. This system
also boasts the absence of complex fluid containment
and accurate fitting of pistons. It is for these reasons that
the bimetallic strip will be used as the basis for the solar
tracker.
Our passive solar tracker design incorporates two
bimetallic strips made of aluminium and steel, posi-
tioned on a wooden frame, symmetrically either side of
a central horizontal axis. The bimetallic strips are
shaded so that the strip further from the sun absorbs so-
lar radiation while the other strip remains shaded in a
similar fashion to the design illustrated in Fig. 2. As
the exposed bimetallic strip gets hotter, the aluminium
bends more than the steel due to its higher coefficient
of thermal expansion. This bending causes a maximum
deflection at the strip midpoint and with the attached
mass; an unbalanced moment results, generating move-
ment. This movement towards the sun rotates the solar
panel also held in the frame, ultimately increasing the
solar flux intercepted by the panel.
Several features, separating this design from others,
are present. Some are essential for the correct working
of the system while others make it more efficient.

A feature essential to the correct working of the sys-
tem includes the bimetallic strip bending towards the
direction of the sun, meaning that the resulting
moment rotates the solar tracker towards the sun
rather than away. This results in the heating of the
bimetallic strip positioned on the other side of the
central axis to the position of the sun.

The shades have a two-fold effect––reflecting solar
radiation away from the shaded bimetallic strip and
onto the heated one (thereby increasing its rate of
heating).

The very nature of the design, i.e. two bimetallic
strips working against each other, will always attempt
to locate the sun, even after the sun emerges from
behind a cloud.

Due to the slow response of the bimetallic strip in
respect to other forms of passive tracker (e.g., par-
tially pressured fluid) less damping is required to con-
trol the motion.
3. Computer modelling
At the design stage, a program was developed to
model the performance of the solar tracker. Each stage
of the system from solar heating to reorientation of
the solar panel towards its optimum position is incorpo-
rated. The system can be broken down into different
phases as shown in Fig. 3.
The outlined stages are now described in more detail.
272 M.J. Clifford, D. Eastwood / Solar Energy 77 (2004) 269–280

An error exists between the angle of the sun and the angle of the tracker,
one of two bimetallic strips is now in a position to absorb solar radiation.

Solar heating of bimetallic strip A

and simultaneous cooling of bimetallic strip B

Bending of each bimetallic strip in accordance to its temperature

increase, with respect to the ambient temperature

Alteration in the moment about the central axis is caused by the change in shape of the
bimetallic strips. This moment imbalance is amplified by the inclusion of a centrally
positioned block mass.

The resulting net moment acts against friction at the bearing surface,
inertia forces and is inhibited by viscous damping forces.

New angular position of the solar tracker

requires the process to be repeated.

Fig. 3. Flow chart describing the processes undertaken in the solar tracker program.

3.1. Thermodynamics X dT AL
Q ¼ mcp ð1Þ
ds
The thermodynamic effects on a solar tracker are two where m = mass; cp = specific heat capacity; dTdsAL = rate
simultaneous heating processes. The presence of an error of change of temperature w.r.t. time.
in the tracker angle from that of the sun, will subject one Therefore, from Eq. (1),
bimetallic strip to solar radiation, while the other is
dT AL
receiving no radiation in the shade. In an environment QSOLAR  QLOSSES ¼ mcp ð2Þ
heated by solar radiation, a material attempts to reach ds
an equilibrium temperature where the energy sources Rearranging Eq. (2), and stepping over small incre-
arriving at the material equal the energy sinks leaving ments of time, a change in temperature of the bar is
it, i.e. found:
P
QDs
Solar radiation gains ¼ Radiation lossesþ DT AL ¼ ð3Þ
mcp
Conductive and convection losses
The resulting increase in temperature can be used to
However, until the material reaches this equilibrium, the model the bending of this composite bar.
difference in energy is conducted through the bar.
3.2. Beam bending
Conduction ¼ Solar radiation  Total losses
As the temperature of the bar increases, the losses Using this result for stepped temperature increase,
become greater and hence the rate of energy conduc- the subsequent bending for both bimetallic strips can
tion decreases, i.e., unsteady heat conduction. How- be found. The deflection of the bar is calculated in two
ever, if the bar is sufficiently thin this complex model stages:
can be simplified to a lumped mass system where it is as-
sumed that the material heats up equally across its (i) The deflection is found for a given temperature
thickness. increase, free of external loading and hence, the
This allows a model for heat conduction through the resulting bending moment, due to heating alone,
bar to be written as is found.
 M.J. Clifford, D. Eastwood / Solar Energy 77 (2004) 269–280 273

(ii) Onto this bending moment, are imposed other bd 3

applied forces like the self-weight of the bar and I AL ¼ þ Ah2 ð12Þ
12
the block masses. where h = distance of bar mid point to neutral axis;
b = width of bar; d = thickness of bar.
(i) The deflection due to a temperature increase is found It is notable that the midpoints of both aluminium
as follows: and steel are located on the tensile side of the neutral
The thermal strains can be calculated using the coef- axis. This is consistent with prior analysis as both bars
ficients of thermal expansion, a, for both materials (Ti- are expanding with increasing temperature.
pler, 1999): Using the above equations, the applied moment, M0,
eThermal ¼ aMaterial LDT ð4Þ on the bar due to thermal effects is found using simple
6
beam bending equations:
where a, for the two materials are aST = 11 · 10 m/K;
EI
aAL = 23 · 106 m/K; L = original length (m); DT = M0 ¼ ð13Þ
temperature change (K) from Eq. (3). R
Assuming the bimetallic strip bends in the shape of where R = average of the two radii of curvature;
an arc of a circle, the following conditions apply: E = YoungÕs modulus; I = second moment of area found
from Eq. (12).
LST þ LeST ¼ RST / ð5Þ (ii) The second stage involves finding the total deflection
considering the other forces present.
LAL þ LeAL ¼ RAL / ð6Þ
The other forces present include
where R = radius of curvature; / = angle of curvature.
But since the angle of curvature, /, is the same
Self-weight of the bar. This is modelled as uniformly
for both materials, Eqs. (5) and (6) can be combined distributed load across the length of the beam and
to give varies as the solar tracker reorientates itself through-
LAL þ LAL eAL LST þ LST eST out the day. The effect of the moment rises to a peak
¼ ð7Þ in the centre of the bar and is given by Eq. (14):
RAL RST
And the radius of curvature for the aluminium can be M WEIGHT ¼ 1=4x2 g cos hðV AL qAL þ V ST qST Þ ð14Þ
calculated as where x = distance along bimetallic strip; g = acceler-
t þ t  ation due to gravity; h = solar tracker angle; V = vol-
AL ST
RAL ¼ RST þ ð8Þ ume of respective material; q = density of respective
2
material.
where t = thickness of bar.

Moment of block mass. The steel blocks are posi-
Combining Eqs. (7) and (8), the following expression
tioned at the midpoint of the bimetallic strip and
for the steel stripÕs radius of curvature can be found:
  are assumed to act as a point load. The bending
LST þ LST eST tAL þ tST  moment again rises to a peak in the centre of the
RST ¼
ðLAL eAL  LST eST Þ þ ðLAL  LST Þ 2 bar and varies with the angle of the solar tracker.
ð9Þ The moment is given by Eq. (15).
The angle of curvature, /, can now be found M POINT ¼ 1=4x cos hgV BLOCK qBLOCK ð15Þ
LAL þ eAL LAL
/¼ ð10Þ
RAL Summing the moments calculated in at stepped inter-
However, the bending moment on a bimetallic strip can- vals along the bar. The net moment applied is found by
not be assumed to be equivalent to a simple applied summing Eqs. (13)–(15):
load, without one other consideration––the shifted neu- Net moment ¼ M THERMAL þ M WEIGHT þ M POINT ð16Þ
tral axis. The neutral axis, i.e. the line that remains the
same length irrespective of the degree of applied bend- By considering the deflection of the beam under this
ing, is necessary to accurately model the flexural rigidity irregular compilation of moments as a boundary value
provided by each bar in the bimetallic strip. Using the problem, a finite difference method can be incorporated
result in Eq. (10), the radius of the neutral axis of bend- into the computer program.
ing is found by Splitting the beam into many small steps, the second
Unstrained length derivative of y with respect to x (i.e., the rate of change
RNEUTRAL ¼ ð11Þ of slope) is given by
/
Hence, the second moment of area for each bar can be d2 y y i1  2y i þ y iþ1
¼ ð17Þ
calculated: dx2 Dx2
274 M.J. Clifford, D. Eastwood / Solar Energy 77 (2004) 269–280
   
And using the standardised equation for bending 1   1  
hi ¼ hi1 þ hi hi1 Ds þ hi h i1 ðDsÞ2 ð25Þ
moment: 2 2
d2 y MðxÞ
¼ ¼ CðxÞ ð18Þ
dx2 EIðxÞ 3.4. Efficiency
A value for the deflection at any point along the
Efficiency is defined as the ratio of power output to
beam is established:
incident radiation. The power obtained by the solar
1 tracker is assumed to be inversely proportional to the
y i ¼ ðDx2 Cðxi Þ þ y iþ1 þ y i1 Þ ð19Þ
2 obliquity factor (Wieder, 1982), i.e. the cosine of the
error angle between the sun and the solar panel. Obvi-
3.3. System orientation ously as this angle increases, the power obtained, and
hence the efficiency, of the solar panel is reduced. Beyond
The torque about the central axis of rotation is now an angle of 90, solar radiation no longer has any direct
unbalanced. This torque, provided by each bimetallic contact on the solar panel and hence no energy conver-
strip is sion occurs, assuming that the diffuse sky component is
negligible and can be ignored for the purpose of these cal-
Torque ¼ gV BLOCK qBLOCK Y MAX ð20Þ culations. For the model in question, in order to find the
efficiency, some factors have to be noted:
Inhibiting this movement is friction at the pivot:
Friction ¼ gMlr ð21Þ
The solar tracker is only able to follow the sun up to
60 either side of vertical.
l = coefficient of friction; M = total movable of mass;
The solar irradiance varies throughout the course of
r = radius of friction moment. a day. It is modelled by an approximation given by
When starting from stationary, the torque resulting Robinson (1964), and shown in Eq. (26):
from the difference in deflections from each bimetallic
strip produce the instigation for movement, and hence Solar irradiance ¼ 1200 sin b W=m2 ð26Þ
reorientation, must over come the friction of the pivot.
This is found by combining Eqs. (20) and (21): where b = solar altitude (measured from the horizon-
tal East).
Net torque ¼ TorqueBMS1  TorqueBMS2

The ambient temperature also increases to a peak

 Friction  ðInertia  hi Þ throughout the day. This temperature rise is mod-

elled by and was derived from known initial bound-
 ðDamping  hi Þ ð22Þ ary conditions:
Ambient temperature ¼ 19:2 sinðbÞ þ 278:9 K ð27Þ
Although initial values of angular velocity and accel-
eration are zero, values during following time steps are These variables dictate the theoretical attainable effi-
not. Using this value for net torque, the angular acceler- ciency, and changing these variables for different
ation can be calculated by considering the total inertia of locations would also alter the maximum possible
the system about its central pivot: efficiency.
Angular acceleration:
 Net torque Three different levels of daily power obtained were
h¼ ð23Þ calculated. These levels allow for the calculation of
Inertia
two efficiency increases––a theoretical optimum (for
Hence, values for angular velocity and, ultimately, angu- the specified conditions) and an actual efficiency pro-
lar displacement of the solar tracker can be found. The duced by the solar tracker.
value for angular velocity is dependent on both the pre- First, it is necessary to find the power obtained by a
vious velocity and the effect of acceleration over the pre- stationary horizontal solar panel as this gives an accu-
vious time period: rate gauge of increases due to any tracking system. This
  is found as follows:
  1  
hi ¼ hi1 þ
2
hi þh i1 Ds ð24Þ X Solar radiation
PowerFIXED ¼  cosðb  hÞ  Ds
Ds
In a similar way, the angular displacement can be ð28Þ
found by considering the effect both the velocity and
acceleration have had on the previous value of But since the angle of the solar panel, h, is fixed at
displacement. 90, i.e. fixed horizontally, this can be simplified to
 M.J. Clifford, D. Eastwood / Solar Energy 77 (2004) 269–280 275

X Solar radiation  p 4. Tracker sensitivity

PowerFIXED ¼  cos b   Ds
Ds 2
X Solar radiation In this section, the effects of varying the dimensions
¼  sinðbÞ  Ds of the bimetallic strips, the damping, and external distur-
Ds
bance such as wind are modelled.
ð29Þ

This value can now be used as a base value against
which others can be measured. The power output of
the solar panel when mounted on the designed solar
tracker is found in a similar way. However, in this case
the angle of the solar panel, h, varies with the tracker re-
sponse throughout the day, and can, over small time
steps, be described using Eq. (30):
X Solar radiation
PowerTRACKED ¼
Ds
 cosðb  hÞ  Ds ð30Þ
However, if the absolute angle jb  hj is greater than
90, the solar panel will receive no direct solar radiation
and 0 W/m2 is noted for that time period. Although this
decreases efficiency, this occurrence is most common in
the earlier morning while the solar tracker is still point-
ing in the opposite direction from the previous evening.
The power output from the sun in the early morning is
much lower than at midday, say, and so the reduction
in efficiency is not as great as first thought. These effi-
ciencies will be used to quantify that loss.
The third level of obtained power, will be that of the
perfect solar tracker. This occurs when the angle be-
tween the solar panel and the sun, i.e. the error, is zero
at all points throughout the day. A value for this is de-
scribed as follows:
X Solar radiation
PowerOPTIMAL ¼  Ds ð31Þ
Ds
By considering how much better the tracked power
output is from the stationary panel output, the efficiency
of the solar tracker can be calculated from Eqs. (29) and
(30):
PowerTRACKED
Extra efficiencyTRACKER ¼ ð32Þ
PowerFIXED
4.1. Bimetallic strip dimensions
Ultimately the dimensions of the bimetallic strips will
dictate the amount of bending at different temperatures
over different time periods. All beam dimensions,
although especially bar thickness, must reach a compro-
mise to satisfy as many requirements as possible––reduc-
tion in bar thickness will allow increased heat exchange,
and hence bending, which accelerates the responsiveness
of the solar tracker. However, thinner strips will also
cause a larger undesirable deflection under the point
mass, thereby increasing the early morning Ôwake upÕ
response time.
4.2. Damping effects
In a system where two opposing means of movement
are constantly working against one another, a level of
damping is required. However, if damping is not applied
in the correct measures, the system will continually oscil-
late or be too sluggish to respond.
In order to maximise the efficiency of the solar panel,
the damping required must be linked to the rate of re-
sponse of the system, i.e. the bimetallic strips. As the flex-
ural rigidity of the bimetallic strip increases, i.e. its
willingness to bend decreases, and the level of system
damping must rise to slow any movement long
enough for the slower system to react. Fig. 5, pro-
duced from the computer model of the system, shows
how the efficiency of the solar tracker changes as the
level of system damping increases. It can be seen that
for an under damped system, excess oscillations actu-
ally decrease solar efficiency. It is not until a critical
level of damping has been reached that the solar
tracker is able to properly function. Effectively, the heavy
levels of damping are necessary to inhibit tracker move-
ment before the bimetallic strips have had time to react.

This value can be compared to the optimum efficiency,
i.e. how much extra power the perfect solar tracker is
capable of obtaining, found by considering Eqs. (29)
and (31):
PowerOPTIMUM
Extra efficiencyOPTIMUM ¼
PowerFIXED
These extra efficiency calculations can be used to test the
impact of a number of key factors controlling the effi-
cient operation of the solar tracker. They will now be
explored.
4.3. External disturbances
Tracking the sun in an ideal scenario would involve
nothing to alter either the system inputs or outputs.
However, when put into practice, factors may con-
trive to ensure either solar radiation does not reach the
ð33Þ
solar tracker (e.g., clouds) or an additional external
force moves the tracker from its optimum position
(e.g., wind, human interference). In either case it is
important for the solar tracker to be able to ÔfindÕ the
sun at any point in the sky, any time throughout the
day. Again, system responsiveness (bimetallic strip
276 M.J. Clifford, D. Eastwood / Solar Energy 77 (2004) 269–280

materials and dimensions) and system damping both However, with a constantly forcing wind, it was nec-
contribute to the solar trackerÕs ability to readapt after essary to examine two cases:
a disturbance.
Two disturbances will be considered: wind, and par- (i) Wind that inhibited motion in the tracking direc-
tial absence of solar radiation caused by clouds. tion produced a lag in the tracking accuracy. This
indicated that accuracy would improve if damping
4.4. Wind were reduced.
(ii) Wind that enhanced the solar trackerÕs desire to
The drag over the solar tracker comprises of two rotate towards the west produced the opposite
main forms: problem. System accuracy would have been
improved had damping been increased.

Friction drag is present when the panels are close to
horizontal. Improbable as either situation is, it is obviously

Pressure drag is prominent when the tracker is posi- impossible to satisfy both, and a balance between he
tioned closer to vertical. two must be sought. It can be assumed that the original
value of damping used was satisfactory. However, fur-
As an initial model, the coefficient of drag is modelled ther experimental testing may be required to asses the ef-
as a flat plate, the area of which varies with tracker fects of additional loads due to wind vortex loading.
angle, h. The seventh power law, as shown in Eq. (34),
can approximate the air velocity over the tracker (White, 4.5. Partial absence of solar radiation
1999):
y 1=7 In the case of a cloud temporarily blocking the radi-
u
¼ ð34Þ ation getting to the solar tracker, the position will not
U0 d immediately change. Since the system is balanced about
where u = fluid velocity at height y; U0 = fluid velocity at its central pivot, the temperature of both bimetallic
boundary layer; y = height above surface; d = height of strips will decrease towards ambient temperature. With
boundary layer (taken from ground). the absence of any applied moment from the effect of
Taking small changes, heated bimetallic strips acting, the system will slow.
However, the moment due to the effect of gravity on
U0 point masses may remain, continuing to enforce a Ôresid-
du ¼ ðdyÞ1=7 ð35Þ
d1=7 ualÕ net moment about the pivot.
In this case, again, a balance must be sought with the
For a known velocity at any height above the ground level of damping applied. Although, heavy damping will
the drag on a tilted plate can be found: inhibit any residual movement, keeping the tracker in a
1 similar position for when the sun is again visible, light
Drag ¼ C d qAU 2 ð36Þ damping will reduce the adjustment time needed to re-
2
focus on the sun.
Using Eqs. (35) and (36), wind drag must be inte- In general with any external disturbance, a quick re-
grated over the area of the tilted solar tracker. Eq. alignment with the sun is desirable (and obtained by
(37) shows the resulting moment: reducing levels of damping), however, to avoid excess
oscillatory movement during normal operation, levels
 2 "Z h1 1=7
1 U0 7=2 of damping must also be sufficiently high. Hence for
dM ¼ C d qA cos h 1=7 ðy  h2 Þ dy each solar tracker design, assigning a value of system
2 d h2
damping to satisfy both objectives as far as possible, is
Z h2 1=7 #2
7=2 fundamental to the accuracy of the overall system.
 ðh2  yÞ dy ð37Þ
h3

where h1 and h3 = heights of each side of the solar 5. Prototype manufacture

tracker; h2 = height of centre axis of rotation.
This moment differential about the central axis, due
to the effect of wind, can be incorporated into the com-
puter model as a continual force or in sporadic burst.
As expected, in the case of sporadic impulses the ef-
fects of wind are least pronounced with higher damping
values. The high levels of system damping facilitate the
dissipation of wind energy.
Once suitable design variables had been chosen from
the results of the computer modelling, a prototype solar
tracker was constructed (Fig. 4).
Material selection was a compromise between two
driving factors––the need to obtain the best material
for the function of the component over its full life, and
the desire to ensure these materials were obtainable in
 M.J. Clifford, D. Eastwood / Solar Energy 77 (2004) 269–280 277

Fig. 4. Prototype of solar tracker using two bimetallic strips.

Ratio of tracking efficiency to fixed pane

1.4

1.2

1.0
efficiency

0.8

0.6

0.4

0.2

0.0
0.0 1.0 2.0 3.0 4.0 5.0 6.0
Damping/ Nm/(rad/s)

Fig. 5. Graph showing the increase in efficiency compared to a fixed solar panel, with increasing levels of damping.

the developing world, cheap, compliant to basic machin-
The materials used for the bimetallic strips were a
ing processes and maintainable. crucial choice. These materials, as well as satisfying
the practical difficulties of acquisition, should also

The fixed frame was constructed from timber, have the following desirable properties for use in a
although any lightweight wood could be used. It bimetallic strip:
may be desirable to change the form of this base – Sufficient difference in coefficients of thermal
depending on method of mounting. If the base is to expansion facilitated the temperature, and hence
be driven into the ground, a stiffer, stronger material stress, differential to be set-up and a bending
may wish to be considered. moment to result.

The moveable frame holding the solar cells and track- – A high yield stress will allow large bending deflec-
ing controls was also constructed from timber. Again tions without permanent strain.
this was chosen on both structural and economic – With continuous heating and cooling, therefore
grounds. Its high specific stiffness, is beneficial to tensile stressing and relaxation, the fatigue resist-
the tracker operation and its ease of fabrication and ance of the material is a vital element in the pro-
acquisition make it an attractive choice. longing of the life of the overall life.
278 M.J. Clifford, D. Eastwood / Solar Energy 77 (2004) 269–280

– The resistance of the material to creep relaxation Thermal Radiation

during use is essential. Stress relaxation would
cause the strain differential generated during heat-
ing to be dissipated, reducing the resultant bending
moment. However, providing the temperature of
the bimetallic strip is less than 30% of the material
melting point, creep failure is not a problem.
– Material stiffness will dictate flexural rigidity of
the bimetallic strip. This is a comprise between a
desire for bending during heating and an aversion Dial
Fixed Support
from it in the absence of solar radiation. Gauge

Considering all of the above factors, a steel–alumin-

ium composite bar was chosen for its ease of manufac-
ture, acquisition, and suitable thermal and mechanical Fig. 6. Experimental set-up for bimetallic strip testing.
properties. General mild steel and aluminium 6082T6/
HE30 were used to construct the prototype.
Experimental conditions were noted and included in

Criteria for the remainder of material selection deci- the computer model:
sions were a balance between ease of acquisition, low
cost, lightweight and reasonable mechanical properties.
Thermal radiation level was measured using a pyro-
nometer and voltmeter.
The completed prototype is shown in Fig. 4.
Ambient temperature was measured using a
thermometer.

Bimetallic strip temperature was measure using a
6. Solar tracker testing thermocouple.

In order to confirm the validity of the proposed the-
ory, it was necessary to compare experimental data with
the numerical predictions obtained from the computer
model.
In order to test the thermodynamic and bending
assumptions made while modelling the bimetallic strip,
tests were carried out to measure mid-point deflection
of a clamped strip under various heating conditions.
The solar tracker was held horizontal to eliminate the ef-
fect of the centre weight on bending. Heating was ap-
plied using 500 W lamps at various distances to alter
the intensity. Deflections were measured at regular inter-
vals as shown in Fig. 6.
Fig. 7 compares the experimental results with the
computer model. The agreement is excellent.
6.1. Dynamic Ôearly morningÕ response
The early morning response can be defined as the
length of time between the emergence of the sun (i.e.,
dawn) to the point when the solar tracker has managed
to completely reorient to itself to face the sun. At this
time of day the solar tracker must use low, but steadily
increasing, values of solar radiation. To quantify this
problem, several tests can be modelled on both the com-
puter program and using the prototype:

14.00
Maximium Deflection / mm

12.00

10.00

8.00

6.00

4.00
Deflection (Computer Model)/ mm
2.00 Deflection (Experimental Testing)/ mm

0.00
0 50 100 150 200 250 300 350 400 450 500
Time /sec

Fig. 7. Modelled and actual deflections of the bimetallic strip due to the effects of thermal radiation.
 M.J. Clifford, D. Eastwood / Solar Energy 77 (2004) 269–280
1200
1000
800
Time/ sec 600
400
200
0
0 200
Fig. 8. Time taken for the solar tracker to reorient from West to East comparing the computer model with experimental results.

The time taken until reorientation was measured and
compared with the computer model, as shown in Fig.
8. It can be seen that as the degree of solar radiation
decreases, the time taken to reorient the tracker
grows exponentially, up until the point solar radia-
tion is insufficient to initiate movement.
However, since the computer model considers only
one set of geographic conditions, a general estimation
for the percentage of unexploited power must be found
by other means.

A better illustration of the effect of response times on
the efficiency of the solar panel can be found by con-
sidering the fraction of energy lost during this time.
Using a computer simulation, and increasing the
total solar radiation spectrum throughout the day,
an indication to the necessity of an additional night
return mechanism can be appreciated.
By considering the values of peak solar radiation suf-
ficient to avoid variable efficiency responses, the compu-
ter model suggests that the efficiency gain provided by a
night return mechanism is slight around 2% extra gain.
This bears significantly on further design considerations,
as the type of geographical location considered has been
that of areas subjected to intense solar radiation, indi-
cating that the increased efficiencies of using a night re-
turn mechanism will only be marginal.
Further manipulation of variables such as damping,
bimetallic strip dimensions and size of point mass have
allowed efficiencies of up to 23% extra to that of fixed
panels.
7. Conclusions and discussion
It can be seen that the benefits of tracking the sun are
substantial. The designed solar tracker has the potential
to increase solar panel efficiency by up to 23%. This de-
279
Time Taken (Computer Model)
Time Taken (Experimental)
400 600 800 1000 1200 1400 1600
Solar Radiation/ W/m 2
sign meets its criteria by adopting a passive method of
control––not draining power from the actual solar cell
for movement. Although solar radiation may be insuffi-
cient in certain geographic locations, an adapted tilted/
dual axis system should provide accurate tracking in
many locations other than those at the equator. Both
the materials used and the manufacturing processes em-
ployed are available in the developing world, allowing
this product to be used, replicated and maintained in
many areas throughout the world.
Although tests displayed a considerable increase in
solar panel efficiency throughout the average day, the
system is capable of improvement. Some of the follow-
ing suggestions allow for future development.
7.1. Night return mechanism
As described in Section 6, the early morning response
of the system, no matter how good, is finite and hence is
a feature that, if changed, will immediately improve effi-
ciency of the solar panel (although more so in areas sub-
jected to lower amounts of solar radiation). As an
additional feature, it should perform its duty after dusk
and before dawn affecting the overall workings of the so-
lar tracker throughout the day as little as possible. One
feasible solution is to use a third bimetallic strip (of the
same or similar materials) to act as a return system. This
would work on the basis that the nights would bring a
considerable temperature drop. The fall in temperature
would lead to a differential rate of thermal contraction
and again produce bending. This activation could then
be utilised to produce a torque to return the system to
its eastern starting point.
7.2. Manually tilted axis
Although the original concept was designed for geo-
graphic locations within 10–20 of the equator, this
should not constrict the systemÕs applicability to that
280 M.J. Clifford, D. Eastwood / Solar Energy 77 (2004) 269–280
region. Providing a sufficient level of solar radiation
strikes the area, this concept will function in many geo-
graphic locations, many of which are greater than 20
from the equator. For example some locations may in-
clude Australasian countries, USA or Mediterranean
countries.
However, using the current solar tracker design
would not be the most efficient set-up for a solar panel.
The option of a non-horizontal axis of rotation would
allow the user to control the tilt of the solar panel, com-
pensating for the sunÕs zenith angle at increased lati-
tudes. Even deciding on tilt angle is not a complicated
procedure––an inclination close to the latitude of the site
will maximise the efficiency of the panel throughout the
year. This increase in efficiency in areas remote from the
equator allows considerable scope for the development
of an improved tracker and certainly merits further
investigation.
7.3. Dual axis system
The next logical improvement from the tilted axis solar
tracker is to use a dual axis system. However, increasing
the number of dimensions in which the solar panel can
be moved is a bold venture. Not only must a second set
of bimetallic strips be designed but also a second degree
of damping must accompany them. Although the instiga-
tion for movement about the two different axes works on
the same principle, the restraints on each are considerably
different. Reorientation for the change in solar azimuth
must be much more responsive, than reorientation to
compensate for the changing solar altitude. Other prob-
lems also generally arise with the increased complexity
of a system such as higher maintenance requirements
and the increased probability of breakdown.
In terms of increased solar panel efficiency, the dual
axis system, in theory, provides the greatest increase in
efficiency at any location and at any time throughout
the year. Although, the gains are more acute further
from the equator, in most geographic regions it is not
sufficient to outweigh the associated costs. If a location
is found that requires the adjustment of tilt from a hor-
izontal orientation, manual adjustment is a satisfactory
compromise between increased efficiency and tracker
complexity. Providing the tilted axis system is accurately
directed between the summer and winter solstices (or
better still, readjusted for seasonal changes) the gains
will comparable to a dual axis system. Additional exper-
imental testing is required to assess the systemÕs corro-
sion resistance, the effect of wind loading, and wear
properties.
Acknowledgments
The authors acknowledge the help and support of
Dr. Ian Care and Dr. Robin Wilson throughout the de-
sign, construction and testing of the solar tracker. We
also thank the referees for their helpful comments. In
particular, our thanks go to Dr. Ross McCluney for
the considerable time and effort spent on making useful
comments on the initial version of this manuscript.
References
Abouzeid, M., 2001. Use of a reluctance stepper motor for solar
tracking based on a programmable logic array (PLA)
controller. Renewable Energy 23, 551–560.
Baz, A., Sabry, A., Mobarak, A., Morcos, S., 1984. On the
tracking error of a self-contained solar tracking system.
Journal of Solar Energy Engineering––Transactions of the
ASME 106, 416–422.
Galen, C., Stanton, M.L., 2003. Sunny-side up: flower heliot-
ropism as a source of parental environmental effects on
pollen quality and performance in the snow buttercup,
Ranunculus adoneus (Ranunculaceae). American Journal of
Botany 90, 724–729.
Helwa, N.H., Bahgat, A.B.G., El Shafee, A.M.R., El Shenawy,
E.T., 2000a. Maximum collectable solar energy by different
solar tracking systems. Energy Sources 22, 23–34.
Helwa, N.H., Bahgat, A.B.G., El Shafee, A.M.R., El Shenawy,
E.T., 2000b. Computation of the solar energy captured by
different solar tracking systems. Energy Sources 22, 35–44.
Lorenzo, E., Perez, M., Ezpeleta, A., Acedo, J., 2002. Design of
tracking photovoltaic systems with a single vertical axis.
Progress in Photovoltaics 10, 533–543.
Markvart, T., 1994. Solar Electricity. Wiley.
Mccluney, R., 1983. Passive optical solar tracking system.
Applied Optics 22, 3433–3439.
Nuwayhid, R.Y., Mrad, F., Abu-Said, R., 2001. The realization
of a simple solar tracking concentrator for university
research applications. Renewable Energy 24, 207–222.
Patil, J.V., Nayak, J.K., Sundersingh, V.P., 1997. Design, fabri-
cation and preliminary testing of a two-axes solar tracking
system. Reric International Energy Journal 19, 15–23.
Radajewski, W., 1987. Water-driven solar tracking mechanism.
Energy in Agriculture 6, 167–176.
Robinson, N., 1964. Solar Radiation. Elsevier.
Ri, S., Ta, O., 1987. A microprocessor controlled solar
tracking system. Journal of Microcomputer Applications
10, 55–62.
Tipler, P.A., 1999. Physics for Scientists and Engineers, fourth
ed. Freeman Worth p. 634.
White, F.M., 1999. Fluid Mechanics, fourth ed. McGraw-Hill
p. 458.
Wieder, S., 1982. An Introduction to Solar Energy for Scientists
and Engineers. Wiley.