Você está na página 1de 8

HELIOSTAT COST OPTIMIZATION STUDY

Finn von Reeken1, Gerhard Weinrebe2, Thomas Keck2, Markus Balz2


1
Dipl.-Ing (FH), Engineer, schlaich bergermann und partner, sbp sonne gmbh
Schwabstraße 43, 70197 Stuttgart, Germany
2
schlaich bergermann und partner, sbp sonne gmbh, Schwabstraße 43, 70197 Stuttgart, Germany.
a)
Corresponding author: f.reeken@sbp.de

Abstract. This paper presents a methodology for a heliostat cost optimization study. First different variants of small,
medium sized and large heliostats are designed. Then the respective costs, tracking and optical quality are determined.
For the calculation of optical quality a structural model of the heliostat is programmed and analyzed using finite element
software. The costs are determined based on inquiries and from experience with similar structures. Eventually the
levelised electricity costs for a reference power tower plant are calculated. Before each annual simulation run the
heliostat field is optimized. Calculated LCOEs are then used to identify the most suitable option(s). Finally, the
conclusions and findings of this extensive cost study are used to define the concept of a new cost-efficient heliostat called
‘Stellio’.

1. INTRODUCTION
Typically, heliostat fields contribute about 25 % to 50 % to power tower overall investment costs. Therefore
heliostat field optimization and cost reduction is of paramount importance to make power towers economically
viable.
The work described in this paper targets at the identification of the most cost efficient heliostat design and size.
This task is very extensive, if a consistent result shall be obtained, because heliostats of different sizes do also
require different structures; just scaling the same design will not yield a meaningful result. Such a procedure may
result in heliostats of different size and different specific cost, but they will not be optimum heliostats.
Furthermore, total costs are worked out for every size and the respective design, since the effort for assembly,
erection and maintenance depends on size and the corresponding number of heliostats required for a given design
power.
Therefore the results presented are not ‘the optimum heliostat size in general’, but the most cost effective variant
(combination of size and structural/mechanical design) of all variants investigated here. As the optical quality of the
individual variants is not identical, specific costs of the heliostats (measured e.g. in Euro per square meter of
concentrator surface) are not an appropriate figure of merit [1] [2]. Instead, field performance must be considered,
too. To this end optical quality and tracking accuracy have to be factored in. This is accomplished by simulating
heliostat fields and the corresponding receiver/power block subsystem and evaluating the resulting levelised costs of
electricity (LCoE).

2. METHODOLOGY
This work aims at finding the most cost efficient heliostat design and size. As a figure of merit for such an
endeavor, specific investment costs of the heliostats, measured e.g. in Euros per square meter of concentrator
surface, are not sufficient. Instead, levelised costs of electricity (LCoE) are calculated for power tower plants with
heliostats employing the investigated heliostat designs. For the calculation of LCoE the publicly available software
SAM is used.
The procedure can be described in a simplified way as follows:
• Based on existing designs and also on experience with heliostat and parabolic trough design, basic
variants of small, medium sized and large heliostats are designed.
• Specific heliostat costs are calculated for every heliostat size and the respective design. The effort for
assembly, erection and maintenance is estimated for every heliostat based on similar components,
manufacturing and assembly procedures. The latter depend on heliostat type and size. Experience and
values from manufacturing and assembly of parabolic trough collector fields is used.
• Tracking and optical quality are determined. For the calculation of optical quality a structural model of
the heliostat is programmed and analyzed using finite element software. By doing so, deviations of the
reflector surface element normals from their desired design orientation due to dead load and wind are
calculated. The RMS value of these deviations is used as input for SAM.

Eventually SAM is used to calculate LCoE for a reference power tower plant equipped with a heliostat field of
the respective heliostats. This is none for all heliostat types and variants. Before each annual simulation run the
heliostat field is optimized using the SAM built-in field optimization tool. Weather data for Upington is used;
financial parameters and power block efficiency have been provided by our former client Sasol. For all other values
the SAM defaults are used. Calculated LCoEs are then used to identify the most suitable option(s).
In order find the exact optimum configuration of a certain heliostat variant, a number of sub variants are created
where small changes in the metal support structure or driving system are made that result in a different optical
quality (beam or tracking). After calculating the LCoE for each relevant variant (variants with higher specific cost
and lower optical quality than their neighbors are sorted out) the optimal configuration can be identified by the
lowest LCoE. The process is schematically depicted in Fig. 1.

FIGURE 1. Methodology scheme

Note: LCoE found must not be considered as absolute values that can be used for a crucial financial project
evaluation; uncertainties concerning some cost factors (e.g. the cost of mirrors per square meter) are still too high for
such a task. Nevertheless, as these uncertainties apply to all heliostat variants investigated in about the same way,
ranking of the heliostats is not affected. Thus the method employed can be considered well suited to identify the
most cost effective combination of heliostat size and design.

3. DEFINITION OF BASIC VARIANTS


As it is not clear at this point which heliostat size is the most cost effective, a comparatively wide span of sizes
must be investigated to identify the most appropriate size. Still, due to the limited time available, the number of
heliostat sizes investigated in detail is limited to five. This will not yield an exact cost optimum, but reduce the large
spectrum of heliostat sizes possible today to a comparatively small range. Five well-selected sizes are enough to
identify cost differences and hence the most appropriate size range. The exact optimum can then be further
investigated in additional studies considering the findings of this study.
In the following, a small (16.7 m2), a medium (43.3 m2) and large heliostats (108.3 m2; 115.6 m2) with
conventional support structure are examined (T-type with torque tube or box). These versions are varied in support
structure, drive solution, etc. Subsequently, the resulting sub-variants are priced and compared. A selection of basic
heliostat variants is depicted in Fig. 2.
Additionally to the conventional heliostat types, some unconventional heliostat and drive concepts types are
analyzed. Here especially the lager versions are of interest, because the simple T-design can be questioned for large
heliostats due to high required bending stiffness.

FIGURE 2. Selection of analyzed heliostat basic variants

4. COSTING
To get the correct dimension of the heliostat members (especially of drive and metal support structure) the loads
acting on the heliostat must be known as precisely as possible. As the governing external action is wind, wind tunnel
tests are mandatory. It is suggested to perform an extensive wind tunnel study on a detailed model to get the best
possible data [3], however, this cannot be done in such an early development stage. Therefore, generic wind pressure
coefficients are used here. In a later step these pressure coefficients need to be verified by more extensive studies for
the selected heliostat concept.
Each variant is then designed according to South African design codes (SANS) and Eurocode (EC) to identify
the minimum required dimensions of all heliostat members. Beneath the survival, also the operation state needs to be
analyzed. As it is the designation of a heliostat to focus the sun light properly, operation is the governing design case
for many members. Hence, it is very likely that the minimum required member dimensions determined in the
structural analysis with survival loads are not enough to generate a satisfying optical quality.
After checking each variant under survival loads, the variant’s optical quality is calculated using the following
equations:
σ slope = σ shape ² + σ mirror ² + σ temperature ² + σ assembly ²
(1)

( 2σ ) + ( 2σ
2
)
2
σ beam = slope waviness + σ specularity 2
(2)
where:

σ shape deformations from dead load and wind load – calculated by FEM
σ mirror error from contour bending, here: from cold bending – calculated by FEM

σ temperature deformations from temperature – calculated by FEM

σ assembly assembly error

σ waviness error from mirror waviness

σ specularity specularity error

For the analysis, the heliostat models are loaded with operational wind and dead loads using the finite element
method software Sofistik [4]. The RMS value of the deviation of the surface normal of each surface element from its
ideal direction and the corresponding steel masses per heliostat are determined.
Since the basic model is a first educated guess based on experience, each basic variant is varied. Initially the
selected cross sections (thickness / section height or section type) are varied. Furthermore, the height of the trussed
girders or their arrangement is varied. The result of this investigation stage is then a table of sub-variants with
associated optical errors and the steel mass.

4.1 Specific Heliostat Cost


For different heliostats costs per square meter reflective surface are determined for the main parts of heliostat
(foundations, metal support structure, drive and controls, reflectors, and assembly).
Some of the positions defined below are based on work offers and contain costs for overhead and profit (e.g.
drives, reflector panels, etc.). For the remaining positions (e.g. foundations, structural steel or assembly costs) a
percentage of 20 % of the position’s costs has been assumed in order to account for the business need of the heliostat
manufacturer and part suppliers.

Foundation

The costs of foundations strongly depend on the soil conditions of the particular site and on the chosen
foundation type. In this study a pile foundation where the pedestal is driven into the ground is considered. As the
soil in Upington is very hard, percussion drilling is recommended by South African foundation specialists. The
drilling diameter is close to the diameter of the pylon section. The remaining diameter difference is then grouted.
The foundation costs depend on the diameter of the pedestal. As the different sub-variants within one heliostat
size class utilizing varying pedestal diameters and wall-thicknesses, there is not only one cost value per heliostat
size. Depending on the number of sub-variants with different pedestal dimensions, the specific costs show a high
scattering. However, as the total costs have to be divided by the reflective surface of the heliostat, the trend goes to
lower cost with increasing heliostat size.

Metal support structure

The structure of glass metal heliostats usually consists of a pedestal and a concentrator supporting structure. The
pedestal consists of a galvanized steel hollow section. The concentrator supporting structure can be built as torque
tube or truss structure. Due to the high number of heliostats required in the field, small heliostats often use a high
content of serial production elements as stamped profiles, while for the medium and large heliostats rectangular
hollow sections (purlins and girders) are used to reach the higher required stiff-ness.
Based on the knowledge of parabolic trough and dish developments, the unit price for light steel structures is
estimated for truss from open sections, truss from rectangular hollow sections, and rolled sections or larger hollow
sections.
Drives and control

For conventional glass/steel heliostats drive costs have to be distinguished between conventional azimuth
elevation (Az-El) heliostat and the so-called slope drive (SD) heliostat configuration [5]. For azimuth elevation
heliostats one slew drive and one linear drive is assumed. The slope drive is characterized by a free axis
configuration (not horizontally or vertically) what allows the use of two linear drives. As linear drives are more cost
efficient than slew drives this drive configuration has significantly lower costs (about 40 to 50% for same tracking
accuracy). In combination with the average lever arm of the linear drive the accuracy of a standard linear actuator
results in a smaller tracking error than reached by conventional slew drives. Drives with different tracking errors and
consequently different costs have been analyzed.
The costs for bearings and limit switches were estimated based on experience from other projects. For the cost
analysis an autonomous heliostat was assumed. The fact of using an autonomous heliostat was considered in the
costs of the controls. These costs include a centralized controller, the local power supplies as well as the power
cabinets.

Reflectors

Costs of the facets are estimated based on inquiries. As the number and position of glass / steel supporting points
is varied, a position for mirror supporting point is considered too.

Assembly

The assembly costs are estimated based on experience values from parabolic trough and dish projects. Included
are the assembly of the sub components into a heliostat structure, the assembling jigs, transport from the assembly
hall to the final position in the heliostat field and the heliostat installation on field.
As the assembly costs depend strongly on the assembling strategy (robotic assembling or high man power,
number of assembly lines) the estimated costs are subject to a greater uncertainty. Since this uncertainty is similar
for all variants analyzed, this is a justifiable approach.
The calculated costs are varying slightly for each variant even though the size is equal. This can be explained
with the difference of optical quality and therefore different field size required for a certain energy output. The
bigger the field, the lower the specific costs of the assembling jigs. Another reason is the higher installation effort
assumed for the slope drive variants.

Summary

The summarized specific heliostat costs range from about 100 €/m² to 170 €/m². As an example, specific costs
for a selection of small heliostats are given in Fig. 3. The first two signs in the variant name indicate the basic
variant (here S1 to S3), the third sign is the running number of a sub-variant. Additional signs x or SD indicate the
selected drive configuration. Here, ‘x’ stands for a low-cost Az-El drive with significantly lower cost but higher
tracking error. ‘SD’ stands for slope drive configuration.
FIGURE 3. Selection of analyzed heliostat basic variants

5. SIMULATIONS AND FINDINGS


The previously identified costs cannot be directly compared, because only the specific costs of materials and
labor is included, the optical efficiency and thus the required solar field size corresponding to a given energy output
is not considered. To determine the relationship between optical efficiency and solar field size, the NREL System
Adviser Model (SAM) was used [6].
Further SAM was used in order to optimize the heliostat field size as well as tower and receiver size for each sub
variant. The optimization criterion therefore is the LCoE that should be as low as possible. As example, the
calculated LCoE for a selection of small heliostats is given in Fig. 4.

20

18 16.7
LCOE real [¢/kWh]

16.3 16.3 16.4 16.2


16.1 16.1 16.1 15.9 15.9 15.9
15.3 15.5 15.3
16

14

12

10
S1-1_x

S1-1_SD
S1-1

S1-2

S1-4

S1-3

S2-1

S2-2

S2-4

S2-6

S2-7

S3-3

S3-6

S3-1

FIGURE 4. Calculated LCoE for selected variants

Since the optical error of the sub-variants of the conventional small torque tube heliostat (S1) is decreasing while
the specific heliostat costs increase, the LCoE of these variants is deviating by 0.3 €-Cent/kWh only. Hence, both
effects are offsetting each other largely. The variant with alternative Az-El drive (S1-1_x) shows equal energy costs
as the corresponding regular variant (S1-1) even though the specific heliostat costs are very different. This example
shows quite well that it is often misleading to compare the specific costs only. The alternative drive has a significant
higher tracking error what consequently leads to a larger receiver, a larger solar field and a higher tower in order to
fulfill the specified requirements.
Due to its lower specific heliostat costs at identical optical error, significant cost reductions are reached with the
alternative slope drive variant (S1-1_SD). The LCoE of the alternative torque tube design (S2) is between
15.9 ¢/kWh and 16.7 ¢/kWh. The most cost effective variant is therefore slightly more economic than the most
favorable variant of S1.
The space frame variants (S3) were especially designed for slope drive, therefore no Az-El drive is analyzed
here. The variant with the lowest LCoE (15.3 ¢/kWh) corresponds to the slope drive variant S1-1_SD.
Bottom line, the slope drive variants provide the lowest LCoEs.

The following general conclusions and findings can be drawn from the results of the investigation:
• It cannot be said in general that a higher optical efficiency consequently results in lower energy costs.
Rather, it is an interaction of structural costs, optical drive costs and optical error. In this analysis, the
medium sized heliostats achieve the most advantageous combination and therefore the lowest energy
costs of all variants analyzed. Due to the high specific drive costs, small heliostats are characterized by
highest LCoE.
• Due to lower costs of linear drives as compared to slew drives, and the potentially lower tracking error,
slope drive versions constantly result in lowest LCOE values. As the free axis arrangement of the slope
drive leads to some additional requirements, the examined T-type heliostats are not purposeful for the
slope drive. During tracking, the heliostat rotates around its surface normal (comparable to pitch and
roll drive configuration). Hence the torque tube would not stay in its horizontal axis. The cantilevers
would be loaded perpendicular to their strong axis (by dead weight). To avoid the corners of the
heliostat touching the ground, the pedestal height needs to be slightly increased. However, the low
LCoE (that consider increased steel mass for longer pedestal etc.) justifies the slightly higher structural
effort.
• It is important to optimize the position and number of supporting points per facet. In the examined case
it is advantageous to increase the number of support points from 16 to 20. However, the ratio of optical
error and solar yield is non-linear, thus with a low optical error as initial value and a reduction of the
error by the amount of X, a lesser increase in yield is achieved than with a larger optical error as start
value. Therefore, a heliostat with lower optical error might react differently to the tested version.
• Due to astigmatism and the consequential spillage losses, very large concentrators and aspect ratios
higher than 1.4 should be avoided.
• A general problem of torque tube heliostats is that the dead and wind loads acting on the reflector
panels and steel structure are collected by purlins and cantilevers and then induced in a central torque
tube. Following the nature of physics, the highest moments appear only on a very small part of the
torque tube (in the centre). The remaining part cannot be reduced in cross-section, thus it has a very low
structural utilization but a high mass. As indicated by the name, a torque tube is good to transfer
torsional moments but has a low bending stiffness to section mass ratio. Thus the dimensions of the
torque tube needed to generate the required stiffness need to be relatively high compared to trussed
structures. Another problem of such a centralized element is that a deflection of the torque tube element
leads to a rotation of the whole concentrator structure.

6. IMPLEMETATION OF FINDINGS
Based on the conclusions of this work a new design has been developed by the consortium formed by schlaich
bergermann und partner, Ingemetal Solar und Masermic - the Stellio heliostat.
• The lowest LCoE could be generated by medium sized heliostats. Thus the Stellio heliostat has a
heliostat surface of 47.5 m².
• In order to avoid expensive and inaccurate slew drives, the slope drive (two linear actuators) has been
selected for the Stellio.
o A pentagonal concentrator shape solves many of the named problems with rectangular
heliostat shapes in combination with the slope drive.
o The axial-symmetric structure provides stiffness where it is needed.
o As trussed members are used, the stiffness to mass ratio is high.
o The axial-symmetric structure allows a good attachment of the drives.
o The nearly roundish shape results in the shortest possible pedestal length.
o Obviously the aspect ratio is one.
• Optimized number and position of supporting points.

A prototype of the Stellio heliostat has been built and is now being operated and tested successfully at the PSA,
Spain (Fig. 5). More information on the Stellio, its development, construction and testing is presented by Balz el al
at the SolarPACES conference 2015 [7].

FIGURE 5. Prototype of Stellio heliostat at the PSA in Spain

ACKNOWLEDGMENTS
The authors want to thank the company Sasol Technology for giving us the opportunity to do this study (and
many others) and to be allowed to use the results after the end of the project for further development steps that
finally resulted in the Stellio heliostat.

REFERENCES
[1] G. Weinrebe, F. von Reeken, M. Wöhrbach, T. Plaz, V. Göcke, and M. Balz, “Towards Holistic Power Tower
System Optimization,” in SolarPACES2013, Las Vegas, NV, USA, 2013.
[2] H. Haberstroh, T. Keck, G. Weinrebe, M. Wöhrbach, C. Husenbeth, F. von Reeken, and M. Balz,
“Optimization of heliostat axes orientations to reduce annual angular ranges,” in SolarPACES 2012, Marrakesh,
Morocco, 2012.
[3] M. Balz and F. von Reeken, “Environmental Loading Conditions for CSP Solar Fields,” in SolarPACES 2014,
Beijing, China.
[4] Sofistik, “SOFiSTiK AG - Finite Elemente & CAD Software für den Ingenieurbau: FEA & CAD Software for
Civil- and Structural Engineering,” Sofistik AG: FEA & CAD Software for Civil- and Structural Engineering,
2015. [Online]. Available: http://www.sofistik.com/index.php?id=2&L=1. [Accessed: 17-Dec-2012].
[5] H. Haberstroh, T. Keck, G. Weinrebe, M. Wöhrbach, C. Husenbeth, F. von Reeken, and M. Balz,
“Optimization of Heliostat Axes Orientations To Reduce Annual Angular Ranges.” SolarPACES, 2012.
[6] NREL, “NREL: System Advisor Model (SAM),” 11-Oct-2011. [Online]. Available:
https://www.nrel.gov/analysis/sam/. [Accessed: 11-Oct-2011].
[7] M. Balz, Göcke, T. Keck, F. von Reeken, G. Weinrebe, and M. Wöhrbach, “Stellio - Development,
Construction and Testing Of A Smart Heliostat,” in SolarPACES 2015, Cape Town, South Africa, October 12-
16.

Você também pode gostar