You are on page 1of 32

Tracking controls for heliostats for

central receiver system

DDP Stage Zero


by

Gaurav Chaudhari (15D170015)

under the guidance of

Prof. Shireesh B. Kedare

Department of Energy Science and Engineering

Indian Institute of Technology Bombay


(11th July 2019)
Abstract

Given higher concentration ratios and higher operational temperature ranges, cen-
tral receiver tower system is very important to study for future of concentrated
solar power technologies. However, various heliostat tracking errors may lead to
poor performance. Therefore, study of these errors and their compensation is cru-
cial. In the following report, sources of errors and methods to contain these errors
are discussed briefly. This report aims to finalize the method of dual axis tracking
for a heliostat and to finalize control logic which can provide corrections to multiple
heliostats simultaneously, for a 1f t2 which will be developed in DDP stage 1 and
DDP stage 2.
Contents

1 Introduction 1
1.1 Solar tracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Objectives and structure of report . . . . . . . . . . . . . . . . . . . 5

2 Literature survey 6
2.1 Heliostat field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.2 Heliostat design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.3 Generalized heliostat tracking geometry . . . . . . . . . . . . . . . . 8
2.3.1 Earth-center coordinate system . . . . . . . . . . . . . . . . 8
2.3.2 Earth-surface coordinate system . . . . . . . . . . . . . . . . 9
2.3.3 Heliostat coordinate system . . . . . . . . . . . . . . . . . . 11
2.4 Heliostat tracking errors . . . . . . . . . . . . . . . . . . . . . . . . 14
2.4.1 Determination of position of sun . . . . . . . . . . . . . . . . 14
2.4.2 Error sources related to heliostat design, components and in-
stallation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.5 Tracking control logic . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.5.1 Open loop control . . . . . . . . . . . . . . . . . . . . . . . . 17
2.5.2 Closed loop control . . . . . . . . . . . . . . . . . . . . . . . 17
2.6 Earlier work on heliostat at IIT Bombay . . . . . . . . . . . . . . . 19
2.6.1 Open loop execution . . . . . . . . . . . . . . . . . . . . . . 19
2.6.2 Feedback mechanism . . . . . . . . . . . . . . . . . . . . . . 19
2.6.2.1 Sushant (2015) . . . . . . . . . . . . . . . . . . . . 19
2.6.2.2 Ravindra (2015) . . . . . . . . . . . . . . . . . . . 20
2.6.2.3 Nishant (2017) . . . . . . . . . . . . . . . . . . . . 20

i
2.7 Gaps in literature . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.8 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

3 Conclusion and future work 23


3.1 Future Tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.1.1 Stage 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.1.2 Stage 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

ii
List of Figures

1.1 Parabolic trough . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2


1.2 Central receiver tower system . . . . . . . . . . . . . . . . . . . . . 3

2.1 Earth-center coordinate system . . . . . . . . . . . . . . . . . . . . 9


2.2 Earth-surface coordinate system . . . . . . . . . . . . . . . . . . . . 10
2.3 Target in earth-surface coordinate system . . . . . . . . . . . . . . . 10
2.4 Heliostat coordinate system . . . . . . . . . . . . . . . . . . . . . . 11
2.5 Orientation of rotation axes and normal vector to reflector surface . 12
2.6 Pedestal tilt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

List of Tables

1.1 Types of solar concentrators . . . . . . . . . . . . . . . . . . . . . . 2


1.2 Various existing heliostat fields . . . . . . . . . . . . . . . . . . . . 4

2.1 Dependency of factors affecting ηO . . . . . . . . . . . . . . . . . . 7


2.2 Various control logic . . . . . . . . . . . . . . . . . . . . . . . . . . 22

iii
Chapter 1

Introduction

Sun is the main source of energy for planet earth. The energy received on earth
from sun is in the form of radiation, known as solar radiation. This radiation can be
used for process heat as well as electricity generation. However, since this radiation
is in a very dilute form, harnessing energy becomes a challenge, especially in the
case of thermal applications requiring higher operation temperatures. To overcome
the dilute nature of the source, solar concentrators are used. There are various
types of solar concentrators, such as central tower receiver, paraboloid dish, fresnel
reflectors, parabolic trough etc.
As earth rotates around sun as well as around earth’s polar axis, the position of
sun varies in the sky with respect to an observer on earth, with day in a year
and time in a day. The position of sun with respect to earth can be calculated
very accurately (+/ − 0.0003 degrees) with various Solar Position Algorithms. For
maximum utilization of incident radiation “tracking” of the motion of sun has to
be incorporated while designing any kind of solar concentrator. Depending on the
type of solar concentrator either dual or single axis tracking is implemented.

1.1 Solar tracking


Depending on type of focus solar concentrators can be categorized as line-focus and
point-focus concentrators as well as continuous and discrete concentrators. All of
the solar concentrators focus on a receiver(s). Depending on the position of receiver

1
Figure 1.1: Parabolic Trough (Olia et al. (2019))

different tracking methods are required. In case of on-axis solar concentrators, such
as paraboloid dish, the receiver lies on the optical axis, hence, the solar concentrator
is oriented such that the optical axis/plane coincides with incident radiation. For
off-axis solar concentrators, the receiver does not lie on the optical axis. Therefore,
a different tracking mechanism has to be implemented. Table 1.1 summarizes the
earlier discussion.

Table 1.1: Types of solar concentrators


Type Line Focus Point Focus On-axis Off-axis No. of tracking axes
Parabolic Trough X X 1
Paraboloid Dish X X 2
Fresnel Reflectors X X 1
Central Receiver Tower X X 2

Both parabolic trough and central tower receiver are the most dominant technolo-
gies. A parabolic trough plant can achieve concentration ratios of up to 100 suns,
a central receiver tower plant can achieve concentration ratios up to 1000 suns (Is-
lam et al. (2018)). Parabolic trough consists of a extended reflecting surface with
cross-section shaped as a parabola (Fig.1.1). Therefore, parabolic trough has a line
focus. The receiver is placed along this line focus. Generally parabolic trough is

2
Figure 1.2: Central receiver tower system (Source: SPIE)

tracked along a single axis. Operational temperature of working fluid varies from
150-350°C (Shahin et al. (2016)). Central receiver tower plant has a receiver placed
on a tower, around which heliostat field is present. As higher concentration ratios
can be achieved, higher temperature ranges of 150-2000°C can be achieved in case
of central receiver tower, which leads to higher thermal efficiency (Shahin et al.
(2016)). For this reason central receiver tower is important to study. Various pilot
as well as commercial power plants based on central receiver tower are established,
which is tabulated in Table 1.2.

3
Table 1.2: Various existing heliostat fields (Gadalla and Saghafifar (2018))
Name Commissioning Capacity Location Scale
year (MW)
CRTF or National Solar Thermal Test Facility (NSTTF) 1979 6 USA Pilot
SSPSCRS (PSA) 1981 0.5 Spain Pilot
Themis solar tower 1983 2.5 France pilot
CESA 1983 7 Spain pilot
Weizmann Institute Of Science 1988 - Israel Pilot
PS10 2006 11 Spain Commercial
Julich Solar Tower 2008 1.5 Germany Pilot
PS20 2009 20 Spain Commercial
BrightSource SEDC 2008 6 Israel Pilot
Eureka 2009 2 Spain Pilot
AORA Solar Tulip Tower Samar 2009 0.1 Israel Pilot
Sierra Sun Tower 2009 5 USA Pilot
Solar Beam Down Plant 2010 0.1 UAE Pilot
Gemasolar 2011 20 Spain Commercial
Acme solar thermal tower 2011 2.5 India Pilot
Coalinga 2011 29 USA Pilot
Lake Cargelligo 2011 3.5 Australia Pilot
Daegu Solar Power Tower 2011 0.2 South Korea Pilot
CTAER variable geometry solar test facility 2012 8 Spain Pilot
AORA Solar Tulip Tower 2012 0.1 Spain Pilot
Yanqing Solar Thermal Power (Dahan Tower Plant) 2012 1 China Pilot
Greenway CSP Mersin Tower Plant 2012 1 Turkey Pilot
Ivanpah 2013 377 USA Commercial
Solugas 2013 4.6 Spain Pilot
Crescent Dunes 2014 110 USA Pilot
Jemalong Solar Thermal Station 2016 1.1 Australia Pilot
Sundrop CSP Project 2016 1.5 Australia Pilot
SunCan Dunhuang 10 MW Phase I 2016 10 China Commercial
Khi Solar One 2016 50 South Africa Commercial

4
1.2 Objectives and structure of report
Objectives of dual degree projects:

1. Design and fabricate a heliostat with reflecting area 1 f t2

2. Develop a heliostat control logic

3. Field testing of heliostat and data gathering

4. Incorporate learning from data gathered into heliostat control logic

Objectives of this report:

1. Finalization of dual axis tracking mode

2. Finalization of heliostat control logic with simultaneous correction to multiple


heliostats

Literature reviewed is described in Chapter 2. Section 2.1 describes various factors


affecting the optical efficiency of heliostat field. Section 2.2 describes two modes
of dual axis tracking. Section 2.3 describes a generalized method for calculating
heliostat orientation angles with arbritrarily oriented axes. Section 2.4 focuses on
various heliostat tracking error sources. Section 2.5 describes various heliostat track-
ing logic. Section 2.6 briefly covers work progress on heliostat at IIT Bombay in
recent few years. Section 2.7 various gaps in literature identified. Section 2.8 is
summary of literature reviewed.
Chapter 3 concludes this report and provides a plan for upcoming 2 stages.

5
Chapter 2

Literature survey

2.1 Heliostat field


A central receiver tower plant mainly consists of heliostat field, receiver tower and
power generation parts. Studies have shown that heliostat field contributes to ap-
proximately 50% of total cost(Piroozmand and Boroushaki (2016); Besarati and
Goswami (2014)). The performance of central receiver plant can be evaluated by
studying the optical and thermal efficiency of overall system. Heliostat field ac-
counts for power losses of 40% (Collado and Guallar (2013)). These losses are
mainly due to various optical factors, such as atmospheric attenuation, spillage
losses etc. The factors affecting the optical efficiency change as per the heliostat
field layout changes. Thus it is crucial to design and optimize heliostat field layout.
Instantaneous optical efficiency (ηO ) becomes a crucial factor while designing helio-
stat field layout. Most of the field layout optimization algorithms break down ηO
into multiplication of various factors (Collado (2009)):

ηO = r × fcos × fatt × fint × fsb (2.1)

where, r is the reflectivity of mirror, fcos is cosine factor, fatt is attenuation factor,
fint is the interception factor and fsb is shadowing and blocking factor.
Table 2.1 summarizes dependency of each factor on heliostat design and tracking,
field layout and externalities. As this report is concerned with heliostat design and
tracking aspect of central receiver tower plant, only fint is discussed, as reflectivity

6
of mirror almost remains same and is only affected by external factors such as lack
of cleanliness.

Table 2.1: Dependency of factors affecting ηO


Factor Heliostat design and tracking Field layout External factors
r X X
fcos X
fatt X X
fint X X
fsb X

fint can be defined as the fraction of reflected sunbeam intercepted by receiver.


Since, reflected sunbeam from a heliostat is not exactly a point it has area. As
heliostat is a off-axis tracking device due to astigmatic aberration this area also
increases in tangential plane, leading to increased chances of spillage, hence reduced
fint . Also, due to heliostat tracking errors, fint is affected.

2.2 Heliostat design


There are various configurations of dual axis tracking (Lipps and Vant-Hull (1978)).
Depending on the orientation of primary and secondary axis most common modes
are as follows:

1. Azimuth-altitude (elevation) tracking : Since it is easy to design, this method


is most commonly used. Azimuth axis, also the primary axis, which is mounted
on ground perpendicularly. Altitude axis is which is parallel to horizon and
perpendicular to azimuth axis is secondary axis.

2. Spinning elevation tracking : The secondary altitude axis is orthogonally


mounted on primary spinning axis pointed to the target.

Astigmatism is defined as the shifting focal length due to oblique incidence of light
on reflector surface. Due to astigmatism, the effective focal length in tangential

7
plane is reduced (Ries and Schubnell (1990)). This phenomenon is known as astig-
matic aberration. As the focal length is reduced the image size increases in tangen-
tial plane.
One way to tackle this problem is having a toroidal/asymmetric (different focal
length in tangential and sagittal planes of heliostat reflecting surface, both planes
are predefined) reflecting surface. As the sun moves in the sky, the tangential and
sagittal planes of image also rotate. Therefore, the toroidal reflecting surface should
be able to rotate such that the tangential and sagittal planes coincide with image’s.
In azimuth-altitude tracking this is impossible since tangential plane of reflecting
surface is fixed to elevation axis, which has constraint for being in a horizontal
plane.
Spinning elevation tracking eliminates this issue, since elevation axis can be rotated
such that the tangential plane of reflecting image coincides with tangential plane
of reflecting surface. A comparison between a azimuth-altitude tracking and spin-
ning elevation tracking was done by Chen et al. (2004). It is shown that spillage
losses by using spinning elevation tracking were reduced by 10-30 % as compared
to azimuth-altitude tracking.

2.3 Generalized heliostat tracking geometry


To obtain generalized tracking formula, one needs to consider 3 coordinate systems.
The method for obtaining heliostat orientation to track sun is described as per Chen
et al. (2006).

2.3.1 Earth-center coordinate system

The origin of this coordinate system is the center of the Earth, C. One axis points
along the polar axis of Earth, CP towards north. The other axis points towards
east, CE. The last one is perpendicular to both, and is given by right hand rule,
CM. Also, the plane passing through CP and CM is called ”meridian plane”.
Figure 2.1 elaborates the coordinate system mentioned:
The position unit vector of the Sun, SˆC , in this coordinate system is defined by

8
Figure 2.1: Earth-center coordinate system (Chen et al. (2006))

following angles:

• δ : The declination angle. The angle made by earth’s equatorial plane with
Sun’s position vector. It is given by the following formula:

• ω : The hour angle.

Therefore,    
S cos δ cos ω
 m  
ˆ
SC =  Se  = − cos δ sin ω 
   
(2.2)
   
Sp sin δ

2.3.2 Earth-surface coordinate system

The origin of the system is the point of interest on the surface of earth, O, which is
also the centre of heliostat reflecting surface. The position of O is defined by the lat-
itude angle(Φ)(positive towards north) in earth-center coordinate system. One axis
points vertically up, that is normal to earth’s surface, called zenith axis, OZ. Other
axis points towards north, ON. The last axis points towards east, OE. Figure 2.2
describes the earth-surface coordinate system as well as the heliostat coordinate
system orientation in earth-surface coordinate system. Now, we have to define SˆC
in earth-surface coordinate system, which can be simply done by multiplying a ro-
tation matrix to SˆC . By rotating SˆC around CE in clockwise direction by angle
Φ we get SˆE which is unit position vector of the Sun in earth-surface coordinate

9
Figure 2.2: Earth-surface coordinate system (Chen et al. (2006))

system.     
S cos Φ 0 sin Φ cos δ cos ω
 z   
SˆE =  Se  =  0 0  − cos δ sin ω  (2.3)
    
1
    
Sn − sin Φ 0 cos Φ sin δ

Now we define unit target position vector, TˆE , in this coordinate system. Figure
2.3 describes the position of target in earth-surface coordinate system.

Figure 2.3: Target in earth-surface coordinate system (Chen et al. (2006))

• φ : The angle made by projection of TˆE on N E plane, with ON axis towards


OE

• λ : The angle made by TˆE with N E plane. For a point above N E plane λ is
negative

10
Therefore,
   
T − sin λ
 z  
ˆ
TE =  Te  =  cos λ sin φ 
   
(2.4)
   
Tn cos λ cos φ

2.3.3 Heliostat coordinate system

Figure 2.4: Heliostat coordinate system (Chen et al. (2006))

Now we define the heliostat coordinate system in this section which shares the
same origin as that of the earth-surface coordinate system, O. The three orthogonal
axes in this coordinate system are OH, OR and OU. The position of these axes is
fixed with respect to the earth-surface coordinate system. The positions/definitions
of these three axes with respect to the earth-surface coordinate system is as follows:

• OH : This axis is defined by two parameters:

– ζ : Rotation angle made by projection of OH on N E plane with ON


towards OE

– α : Angle of OH with N E plane. For point above N E plane, α is


negative

• OU and OR : OU and OR point towards OE and OZ respetively, when


α = 0 and ζ = 0.

11
So the definition of OU and OR depends on orientation of OH. The advantage
of this method is we have freedom over choice of orientation of OH which can
be, for example, a rotation axis, which will lead to some interesting calculations.
Important thing to note here is that the heliostat frame is defined using the original
position of heliostat, when the heliostat oriented in such a way that the normal of
the reflector surface is coincident with OH and rotation axis KK 0 is coincident with
OU . A clear picture of this description is described in Figure 2.5 below.

Figure 2.5: Orientation of rotation axes and normal vector to reflector surface (Chen
et al. (2006))

The unit normal vector NˆH can be defined by following angles:

• ρH : Angle of rotation of NˆH around OH

• θH : Angle of rotation of NˆH around KK 0

Therefore,    
N cos ρH cos βH
 r  
NˆH = Nu  =  sin ρH cos βH  (2.5)
   
   
Nt sin βH
where,
π
βH = − θH (2.6)
2
Now by rotating SˆE and TˆE to this coordinate system to get the normal vector of
heliostat, NˆH can be calculated.

12
     
T cos α 0 sin α 1 0 0 − sin λ
 r     
TˆH = Tu  =  0 − sin ζ  (2.7)
      
1 0  0 cos ζ  cos λ sin φ 
      
Tt − sin α 0 cos α 0 sin ζ cos ζ cos λ cos φ
and
      
Sr cos α 0 sin α 1 0 0 cos Φ 0 sin Φ cos δ cos ω
SˆH =
       
Su  =  0
  
  
1

0  0

cos ζ − sin ζ 



 0

1
 
0   cos δ sin ω 
 
(2.8)
St − sin α 0 cos α 0 sin ζ cos ζ − sin Φ 0 cos Φ sin δ

where, TˆH and SˆH are the unit position vectors of target and Sun in heliostat co-
ordinate system.

Now to find NˆH such that it bisects the angle between SˆH and NˆH it must sat-
isfy the following equation,
1
NˆH = (SˆH + TˆH ) (2.9)
2 cos θi
where,
θi = 0.5 arccos (SˆH · TˆH ) (2.10)

Solving equation 2.9,

 
− sin α(cos Φ cos δ cos ω + sin Φ sin δ) − cos α sin ζ cos δ sin ω+
 
 

 cos α cos ζ(− sin Φ cos δ cos ω + cos Φ sin δ) + sin α sin λ+ 

 
cos α sin ζ cos λ sin φ + cos α cos ζ cos λ cos φ 
 
π 
θH = −arcsin  
2 
 2 cos θi 

 
 
 
 
 

(2.11)
and
 
− cos ζ cos δ sin ω+
 
 
 sin ζ(sin Φ cos δ cos ω − cos Φ sin δ) + cos ζ cos λ sin φ − sin ζ cos λ cos φ 
ρH = arcsin 
 
2 cos θi cos βH

 
 
 

(2.12)

13
Therefore, θH and ρH gives the orientation of heliostat reflector normal for Sun
tracking irrespective of the orientation of heliostat axes. The orientation of axes
will be defined by α and ζ.

2.4 Heliostat tracking errors


Central receiver tower plant needs to overcome various challenges for optimal op-
eration. Heliostat needs to be oriented such that the reflected sunbeam should be
intercepted by receiver. If the heliostat is not properly aligned, the reflected sun-
beam may partially fall on the receiver or miss the receiver entirely. The receiver is
designed for certain flux profile. To maintain this flux profile just aiming heliostats
at receiver is not enough. Each heliostat is needed to aim at a defined region on
the receiver surface. This defined region may also vary with time for each heliostat.
There are various external and inherent design factors that affect the accuracy of
aiming point of heliostat. Inaccuracies in heliostat aim point are called “heliostat
tracking errors”.

2.4.1 Determination of position of sun

Position of sun in the sky with respect to an observer on surface of earth with time.
This position can either be measured using sensors or can be calculated analyt-
ically. However, these methods have margin of errors associated. Solar Position
Algorithm (SPA) by Reda and Andreas (2004) is the most accurate algorithm to
obtain position of sun, with an accuracy of 0.0003°( 0.005 mrad). Since advance-
ment of computational technologies, SPAs can be executed very fast. Also, since
the accuracy is quite acceptable, most of the literature reviewed focuses on reduc-
ing tracking errors associated with heliostat design and various practical issues.
Also, apparent sun position due to scattering and refraction very much depends on
weather conditions, in countries like India the variation in weather is drastic, which
makes the task of dynamic tracking more difficult. In the report Nayak et al. (2015),
one of the reason for low performance of 1 MW solar thermal facility located near

14
New Delhi, India was due to tracking errors. One of the reason for tracking errors
was atmospheric refraction. Due to haziness in sky atmospheric refraction of solar
radiation takes place. This leads to shift in apparent position of sun.

2.4.2 Error sources related to heliostat design, components


and installation

A study by Freeman et al. (2014) states different kind of error sources associated
with heliostat design and installation. In the study a “Risk/Reward analysis” is
done to obtain a “low hanging fruit”, that is, to identify high impact error sources
with least difficulty to eliminate or reduce them. As stated by Freeman, this study
should be used as a guide while designing a heliostat, since the study is not quan-
titative.

1. Pedestal Tilt : Pedestal tilt can be defined as the tilt caused by faulty pedestal
mount. Pedestal mount might be affected by various external factors, wors-
ening pedestal tilt.

Figure 2.6: Pedestal tilt (Freeman et al. (2014))

2. Reference angles : For alignment of heliostat the control logic requires ref-
erence angles. For calculation of change of azimuth angle, the control logic

15
needs to know the accurate reference angle of true north. Due to pedestal tilt
and other causes, the azimuth axis of rotation might not be perfectly verti-
cal, that is, coinciding with zero degree zenith angle which is generally the
reference angle for elevation angle calculation.

3. Gear backlash : Due to free play between gears gear backlash takes place. Var-
ious types of gear mechanism have different backlashes. Generally, planetary
and worm gears have lesser backlash, but they are expensive.

4. Backdrive : Mirror mounts of heliostats can be very heavy. Due to gravity, the
heliostat shifts, causing unwanted rotation in driving motors. This is known
as backdrive.

5. Encoder resolution : Encoders are used to measure the angular rotation of


rotary motors and length in case of linear actuators, which drive heliostat
rotation axes. A lower encoder resolution results in larger margin of error
while orienting heliostat.

6. Boresight error : Also known as canting error, arrises due to deviation of


effective normal of heliostat facets from normal defined for driving axes.

7. Non-orthogonality of rotational axes

The error sources stated above are not independent of each other. So it is very
difficult to predict the pointing errors correctly. However, the effective pointing
error caused by the aforementioned error sources can be compensated either by
calibration or providing the controller with a feedback.

2.5 Tracking control logic


As stated by Baheti and Scott (1980) pointing error should be less than 0.1°( 1.75
mrad) for minimization of spillage. Due to various external factors, such as wind
loading, changes in leveling of ground due to rain, the errors caused by various error
sources also vary. Also, since heliostat tracking is a dynamic system the pointing

16
error also varies with time. To compensate the error a robust control logic has to
be developed. Heliostat control logic can be categorized as open loop control and
closed loop control logic.

2.5.1 Open loop control

Open loop control, as the name suggests is open loop. There is no feedback. Open
loop control is based on mathematical models. The mathematical model may or may
not use calibration for accommodation of errors. Generally, calibration data is used
for mathematical modeling of errors. This is called model based open loop control.
This model is then used for compensation of errors. However, since error sources
can vary over time, a same model cannot be used over time. Hence, calibration
needs to be done after certain interval of time for every heliostat.
State-of-the-art open loop control technique is to use a dummy screen installed just
below the actual receiver and a imaging device facing the dummy screen. A single
heliostat is ordered to aim at the centre of the dummy screen. Image of the dummy
screen is then captured by the imaging device. Using image processing techniques
the actual position of the reflected sunbeam is obtained. This offset is then used
for calibration of the particular heliostat. This is repeated for all heliostats.
Effect of threshold intensity for detection of correct position of reflected beam is
observed by Sushant (2015). It is found that a lower and higher threshold intensity
is to be used during clear and cloudy sky respectively. The amount of correction
required was obtained to be around 0.32 mrad in azimuth and 1.49 mrad is elevation
axis.

2.5.2 Closed loop control

In closed loop control a real time feedback is provided to each heliostat for aim
point correction. Closed loop control system also operates in open loop control
mode for rough orientation of heliostat. Generally a analytical model is used. For
more accurate aim point closed loop control is executed.
Kribus et al. (2004) obtained heliostat tracking accuracy within 0.1 - 0.3 mrad.

17
The method used four cameras surrounding the receiver. Two were located on ei-
ther side of the receiver and two were located above and below the reciever. These
cameras are oriented towards the heliostat field. Each heliostat was identified and
the brightness of each heliostat was compared on cameras on opposite sides of the
receiver. The principle of detection of misalignment is brightness imbalance in op-
posite cameras. This method however is not implemented in a commercial central
tower receiver. This is due to difficulty in operation of cameras near high tempera-
ture receiver.
R. Convery (2011) has designed a lab-scale closed loop control system using piezo-
electric actuators and photo-diodes. In this method, four photo-diodes are installed
around the receiver. Piezoelectric actuators are installed on the reflecting surface
of the heliostat. These actuators vibrate the reflecting surface resulting in modu-
lation of reflected sunbeam. If a heliostat(s) is misaligned, the sunbeam falls on
one (or more) of the photo-diodes. The photo-diode signal is then processed using
Fast Fourier Transform analysis for identification of misaligned heliostats. Since
each heliostat is assigned a unique frequency multiple misaligned heliostats can be
aligned simultaneously.
Bern et al. (2017) describes an approach aimed at the extraction of heliostat focal
spot positions within the receiver domain by means of a camera system looking at
the receiver, away from the rough conditions close to the focal area. The approach
is similar to that of R. Convery (2011) with regards to identifying focal spots us-
ing modulated frequencies, albeit the driving frequency used is much lower (0.2 to
2 Hz instead of the kilohertz audio frequency range used by R. Convery (2011).
The slow movements allow for the utilization of the heliostat actuators to modu-
late the reflected sunlight instead of adding additional hardware, cabling and I/O
infrastructure to each heliostat. This makes the control strategy easy to deploy in
existing plants. The authors did mention that the selection of the driving frequency
should be carefully selected, as wind-induced vibrations and resonant frequencies
of structures may compromise the original driving frequency.

18
2.6 Earlier work on heliostat at IIT Bombay
The miniature model of heliostat is same for all three models studied. The model
is a dual axis model, based on azimuth-altitude mode of tracking.

2.6.1 Open loop execution

The open loop execution is same in case of both Sushant (2015) and Ravindra
(2015). Both of them use ‘Sun Harvester Shield’, which calculates the position
of Sun based on real time and the program hard-coded in the hardware. The position
of target is pre-fed to shield. Using this method open loop is executed. Both models
have a function in the controller for checking if the feedback is available or not. If
there is no feedback, this open loop is directly executed.
Nishant (2017) on the other hand uses Solar Position Algorithm given by Reda
and Andreas (2004). The algorithm accurately calculates position of Sun with an
accuracy of 0.005 mrad.

2.6.2 Feedback mechanism

2.6.2.1 Sushant (2015)

In case of Sushant (2015) the feedback is provided using a camera facing the dummy
screen. The image of the dummy screen with the beam falling on it is captured
by a single camera. Then it is converted to black and white image, by defining
a threshold value of intensity. This threshold is basically, the minimum intensity,
which will be considered as a beam on the screen. As the results concluded, having
a lower value of threshold gives better results in case of a clear sky and the opposite
for cloudy sky.
This conversion of the captured image into a black and white image makes it easier
to find the centroid of the sunbeam. The identification of the centroid is then sent
back to the heliostat controller, for correction.

19
2.6.2.2 Ravindra (2015)

A 1.8x1.8m test receiver is used instead of a dummy receiver. The receiver has four
cameras attached to 4 sides of receiver facing the heliostat field.
The images of the heliostat field are captured by the cameras. For each heliostat
in field a Region of Interest (RoI) is then defined for each camera. This RoI
contains the heliostat concerned. All the other pixels of the image are set to black.
The maximum intensity in region of interest is the then considered for further
calculations. Next step is learning. The sun beam is reflected on the test receiver
using manual controls, along the length and width of the receiver passing through
the center. This is done step by step, to get the intensity ratio and the position
on the screen mapping. It is found out that there is a linear relationship between
the log of maximum intensity ratio of opposite cameras and the position on the
screen in the respective axis. This observation is used to find a regression model.
This regression model is then used to provide real time feedback to the heliostat
controller, which remaps the log of intensity ratio captured by opposite cameras
to position of sunbeam on screen. However, only detection of offset was validated.
The complete closed loop was not demonstrated.

2.6.2.3 Nishant (2017)

The methodology is similar to that of Ravindra (2015), but the learning step is
automated using a code. Closed loop control is also demonstrated.

2.7 Gaps in literature


In case of R. Convery (2011), the photo-diodes can only sense the heliostat which
are misaligned and such that there is spillage. In reality, the heliostats need to
aim at the receiver at certain region of receiver. This is due to the receiver are
designed for certain flux profile. Failing to meet this flux profile creates thermal
stresses in the receiver which is unfavourable. The photo-diodes can only sense the
spilling heliostats, the exact location of the reflected beam cannot be identified on
the receiver, which is very crucial for operation of central receiver tower system.

20
Also, if the reflected sunbeam totally misses the receiver and the photo-diodes, a
feedback cannot be provided.
In the method presented by Sushant (2015), the problem in detection of sunbeam
on the test screen occurred when there was direct sunlight falling on the screen. In
case of non-overlapping, a filter was which removes unwanted objects of size larger
and smaller than predefined size are removed. But this method can fail if there are
sunlight spots on screen of similar size to that of sunbeam from heliostat.
In case of partial overlapping of direct sunlight on the screen with sunbeam, the
model simply cannot work. However, this might not be an issue for a practical cen-
tral receiver tower system, since the concentration ratio of reflected beam is much
higher. Also, this scenario is highly likely to occur in a heliostat field. In this model
the operation has to stop till the partial overlapping is gone. Also, a very important
thing to note here is, this model can only work for a single heliostat at one time.
Four iterations are required to keep errors into acceptable limits.
One clear advantage of both models, Ravindra (2015) and Nishant (2017), over
Sushant (2015) is that both models can provide corrections to multiple heliostats
at the same time. However, the calibration process in both model has limitations.
As clearly mentioned by Nishant (2017), the calibration data varies significantly
within 30 minutes of operation. The time taken for calibration is roughly around
100 seconds. Also, the model by Nishant (2017) requires 4-5 iterations per cycle,
which roughly accounted for 70 seconds.
One important issue with both Ravindra (2015) and Nishant (2017) was the re-
duction in sensible region on the test screen. This was due to the intensity in
circumsolar region around Sun varied exponentially. As the image of the Sun moved
close the cameras, the resolution of position on the screen reduced drastically as
the values of intensity saturated, i.e., intensity observed reached maximum value at-
tainable. As stated by Ravindra (2015), sensible region on the receiver screen(1.8m
x 1.8m) was reduced to 1.2m x 1.2m.
Also, the problems arising from cloud covering in the sky was not addressed by both
Ravindra (2015) and Nishant (2017). In case of Sushant (2015) one interesting in-
sight, the effect of threshold intensity to obtain calibration data was established.

21
However, there is no automatic method which changes this threshold intensity.

2.8 Summary
Various issues arising during operation of central receiver tower system due to choice
of heliostat design has been discussed. Various heliostat control logic are discussed
briefly and few gaps in literature regarding control strategies are identified. A gen-
eralized method for obtaining heliostat orientation angles obtained from literature.
From Table 2.2, we can see that only open loop method of calibration does not

Table 2.2: Various control logic


Closed Loop
Open loop Error
Control Logic (real-time feedback) Multiple heliostats?
(Calibration) (in mrad)
Vision based Sensor based
Kribus (2014) checkmark 0.3 X
Convery (2011) X 1.5 X
Sushant (2015) X 1.49
Bern (2017) X - X
Ravindra (2015) X - X
Nishant (2017) X 0.1 X

provide corrections multiple heliostats simultaneously.

22
Chapter 3

Conclusion and future work

From literature survey, objectives of this report can be justified as follows:

1. Finalization of dual axis tracking mode:


From Sections 2.1 & 2.2 it is clear that spinning elevation tracking has various
benefits over generally used azimuth-altitude tracking. Also, azimuth-altitude
tracking has been demonstrated at IIT Bombay previously.

2. Finalization of heliostat control logic with simultaneous correction to multiple


heliostats:
Open loop calibration method lacks support for simultaneous correction to
multiple heliostats. This problem can be resolved by using piezo-electric ac-
tuators and a grid of photo-diodes.

3.1 Future Tasks

3.1.1 Stage 1

1. Design a spinning elevation heliostat with 1 f t2 mirror area

2. Fabrication heliostat

3. Design and fabricate a photo-diode grid for detection of offset for calibration

4. Develop a control logic

23
3.1.2 Stage 2

1. Testing of fabricated heliostat and photo-diode grid

2. Develop a data collection method to store calibration data which can be in-
corporated in a learning model

24
Bibliography

Baheti, R. and Scott, P. (1980). Design of self-calibrating controllers for heliostats in


a solar power plant. IEEE Transactions on Automatic Control, 25(6):1091–1097.

Bern, G., Schöttl, P., Heimsath, A., and Nitz, P. (2017). Novel imaging closed loop
control strategy for heliostats. In AIP Conference Proceedings, volume 1850, page
030005. AIP Publishing.

Besarati, S. M. and Goswami, D. Y. (2014). A computationally efficient method for


the design of the heliostat field for solar power tower plant. Renewable Energy,
69:226 – 232.

Chen, Y., Kribus, A., Lim, B., Lim, C., Chong, K., Karni, J., Buck, R., Pfahl, A.,
and Bligh, T. (2004). Comparison of two sun tracking methods in the application
of a heliostat field. Journal of Solar Energy Engineering, 126(1):638–644.

Chen, Y., Lim, B. H., and Lim, C. (2006). General sun tracking formula for he-
liostats with arbitrarily oriented axes. Journal of Solar Energy Engineering-
transactions of The Asme - J SOL ENERGY ENG, 128.

Collado, F. J. (2009). Preliminary design of surrounding heliostat fields. Renewable


Energy, 34(5):1359 – 1363.

Collado, F. J. and Guallar, J. (2013). A review of optimized design layouts for


solar power tower plants with campo code. Renewable and Sustainable Energy
Reviews, 20:142 – 154.

Freeman, J., Kiranlal, E., and Rajasree, S. (2014). Study of the errors influencing
heliostats for calibration and control system design. In International Conference

25
on Recent Advances and Innovations in Engineering (ICRAIE-2014), pages 1–8.
IEEE.

Gadalla, M. and Saghafifar, M. (2018). A concise overview of heliostat fields-solar


thermal collectors: Current state of art and future perspective. International
Journal of Energy Research, 42(10):3145–3163.

Islam, M. T., Huda, N., Abdullah, A., and Saidur, R. (2018). A comprehensive
review of state-of-the-art concentrating solar power (csp) technologies: Current
status and research trends. Renewable and Sustainable Energy Reviews, 91:987 –
1018.

Kribus, A., Vishnevetsky, I., Yogev, A., and Rubinov, T. (2004). Closed loop control
of heliostats. Energy, 29(5-6):905–913.

Lipps, F. and Vant-Hull, L. (1978). A cellwise method for the optimization of large
central receiver systems. Solar Energy, 20(6):505–516.

Nayak, J., Kedare, S., Banerjee, R., Bandyopadhyay, S., Desai, N., Paul, S., and
Kapila, A. (2015). A1 mw national solar thermal research cum demonstration
facility at gwalpahari, haryana, india. Current science, 109:1445–1457.

Nishant, K. (2017). Development of tracking logic for heliostat based on feedback


by camera.

Olia, H., Torabi, M., Bahiraei, M., Ahmadi, M. H., Goodarzi, M., and Safaei,
M. R. (2019). Application of nanofluids in thermal performance enhancement of
parabolic trough solar collector: state-of-the-art. Applied Sciences, 9(3):463.

Piroozmand, P. and Boroushaki, M. (2016). A computational method for opti-


mal design of the multi-tower heliostat field considering heliostats interactions.
Energy, 106:240 – 252.

R. Convery, M. (2011). Closed-loop control for power tower heliostats. Proceedings


of SPIE - The International Society for Optical Engineering, 8108:17–.

26
Ravindra, K. (2015). Design and development of self-learning control of heliostat
using feedback.

Reda, I. and Andreas, A. (2004). Solar position algorithm for solar radiation appli-
cations. Solar energy, 76(5):577–589.

Ries, H. and Schubnell, M. (1990). The optics of a two-stage solar furnace. Solar
Energy Materials, 21(2-3):213–217.

Shahin, M. S., Orhan, M. F., and Uygul, F. (2016). Thermodynamic analysis of


parabolic trough and heliostat field solar collectors integrated with a rankine cycle
for cogeneration of electricity and heat. Solar Energy, 136:183 – 196.

Sushant, M. (2015). Design and development of feedback control for heliostat.

27

You might also like