PROGRESS IN PHOTOVOLTAICS: RESEARCH AND APPLICATIONS
Prog. Photovolt: Res. Appl. 2012; 20:356 371
Published online 6 September 2011 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/pip.1136
APPLICATIONS
Third generation EUCLlDES concentrator results
*
Marta Vivar , Ignacio Antón, Dolores Pachón and Gabriel Sala
Instituto de Energía Solar, Universidad Politécnica de Madrid, Ciudad Universitaria, S/N 28040, Madrid, Spain
ABSTRACT
This paper presents salient results from the third generation •UCLID•S linear photovoltaic concentrator, which was devel-
oped within the framework of the ID•OCONT• •uropean project. There were two broad objectives for this project: firstly,
to review and resolve the difficulties associated with previous prototypes (the Madrid prototype and the Tenerife plant),
where the lack of qualification standards and quality control were identified as significant causes of failure under real world
operation; and secondly, to optimise and commercialise the functionally effective components. During the development pe-
riod, research activities focussed on cell encapsulation, the reliability of the system components, and the monitoring of field
performance across three different test sites. The third generation •UCLID•S concentrator system has achieved an overall
efficiency of 10%, which is similar to the efficiencies of previous •UCLID•S systems. The main third generation improve-
ments relate to the industrialisation and improved reliability of single element system component. The expected system
cost, for an annual production of 2 MW, is Ź 3.70/Wp. However, the maximum allowable projected system cost under
the Spanish feed in tariff of 2010 is Ź 3.00/Wp. This will be very difficult to achieve with the current system efficiency.
Copyright © 2011 John Wiley & Sons, Ltd.
KEYWORDS
photovoltaic concentrator; linear; parabolic; low concentration; silicon
*Correspondence
Marta Vivar, Instituto de Energía Solar Universidad Politécnica de Madrid, Ciudad Universitaria, S/N 28040 Madrid, Spain.
E mail: [e-mail]
Received 31 October 2010; Revised 16 March 2011
1. BACKGROUND The first •UCLID•S prototype, operating at 32X, was
designed and established in 1995 at the I•S (Madrid)
The •UCLID•S concentrated photovoltaic (Ä„V) technol- [2 4] with assistance from ZSW (Zentrum für Sonnenenergie
ogy has been under development for 12 years between und Wasserstoff Forschung, Germany). As a result of the
1995 and 2007 at the Instituto de •nergía Solar Universi- satisfactory results achieved, an entire demonstration plant,
dad Ä„olitécnica de Madrid (I•S UÄ„M), with the principal the 480 kWp  •UCLID•S TH•RMI• power plant of
collaboration of BÄ„ Solar [1]. The technology consists of a Tenerife, was proposed and developed in 1998 [5]. This
cylindrical parabolic trough concentrator with silicon solar plant, known as the  •UCLID•S II and operating at 38X,
cells working at low to medium geometric concentration was at the time the largest CÄ„V system in the world. The
and employs a passive cooling arrangement. The optical Instituto Tecnológico y de •nergías Renovables of Tenerife
system consists of asymmetric parabolic mirrors, with the was the principal investor and also the local researcher.
structure rotating about a single north south orientated This plant did not reach the expected field performance,
horizontal axis, with active tracking. The outcome of the and it also revealed some of the problems of industrialising
development on three different prototypes, each corres- the early •UCLID•S technology [6]. In the demonstration
ponding to Research and Technical Development projects plant, medium industrialised components such as the
and demonstration projects, has been summarised in the tracking system, the extruded heat sinks and the support
succeeding paragraphs. The main objective of the third structure performed successfully. The main source of fail-
generation research project was to demonstrate high con- ures lays in the elements that were less commercially
centration photovoltaic (CÄ„V) performance potential and ready: innovative elements such as the mirrors and some
overall cost reduction with respect to conventional flat receiver components. These items had passed a brief qual-
plate Ä„V systems, mainly via significant reduction of ex- ification and type approval prepared for the occasion, but a
pensive cell area. comprehensive quality control programme during the
356 Copyright © 2011 John Wiley & Sons, Ltd.
M. Vivar et al. Third generation EUCLIDES concentrator results
manufacturing processes did not follow. This was due in scientists visiting the site or attending conferences where
part to the lack of specific standards for CÄ„V systems, •UCLID•S II results have been presented. This experi-
which contributed to several subsystem defects being ig- ence must now be built on by developing new designs to
nored. As a consequence, some of the components were determine solutions for two of the critical aspects of the
not subjected to adequate quality control and suffered a system: the receiver design, and the mirror industrialisa-
rapid degradation under real world operation conditions af- tion. By resolving these questions, the real commercial op-
ter field installation. Contributing factors for this rapid deg- portunities for •UCLID•S and similar technologies within
radation, which was not observed in the first prototype in the Ä„V market could be assessed.
Madrid, were the different manufacturing processes along From the previous considerations, the identified research
with poor quality control in comparison with the manufac- areas for the development of a new version of the
turing for the first •UCLID•S prototype, the geographic •UCLID•S system, which also include the generic prob-
conditions of the Tenerife test site, close to the sea and lems of all the low to medium CÄ„V systems, include the
with strong winds, and sulphuric gases due to the volcanic following: improved mechanisms for the control of electri-
nature of the island. As an example of the effect of the sul- cal shorts within the receivers; improved thermal transfer to
phuric gases, the mirrors lost their reflecting properties af- enable more efficient cooling of the solar cells; improved
ter a short period. After a short time (a few days), 20% of electrical interconnection within and between the receivers;
the mirror edges had lost sealing, and water began to enter. improved optical coupling between the mirrors and the
Slowly but progressively, the entire reflective silver layer structure; defining the component qualification tests; de-
was eliminated. signing and implementing the data acquisition system and
With respect to the receiver design, the same encapsula- procedures for monitoring and analysis of the results; and
tion method was used by BÄ„ Solar in the •UCLID•S I pro- a comprehensive identification of the failure modes and
totype (1996) as well as in the •UCLID•S II Tenerife degradation mechanisms that cause efficiency losses.
plant [7]. •UCLID•S I II method consisted of using a The third generation •UCLID•S concentrator [8] was de-
thermal film, which is an electrically insulating film with veloped within the framework of the ID•OCONT• project
or without adhesive in any of its sides. It provides a [9], as funded by the •uropean Commission. The overall
method for glueing, or mechanically fixing, the cells to objective of this project was to identify the most appropri-
the substrate. This provides electrical insulation, with a ate Ä„V concentrator system configuration, based on silicon
voltage stand off of more than 3000 V, and specific ther- concentrator cells, for power plant applications in different
mal resistance is 6°C cm2/W. The problem with this mate- regions. The range of concentration levels, extending from
rial, as became evident in Tenerife, was its change in 2X to over 100X, was explored experimentally, with six
adhesion properties when heated above 100°C. Above this concentrator systems being manufactured according to pre-
temperature, the thermal film loses its adhesive properties vious development concepts by the project partners (ZSW,
and therefore also its heat transport characteristics. Further- University of Ulster, and Universita di Ferrara). Within the
more, this material also revealed poor performance as an project framework, the third generation •UCLID•S con-
electrical insulator in the •UCLID•S II plant. After a centrator was developed for two concentration levels:
few days of operation, several modules lost insulation, pro- 12X and 24X. The purpose of this dual approach was to
ducing short circuits in the arrays, which stopped the in- explore lower gain values with respect to previous
verter and caused failure of the entire system. This failure •UCLID•S systems in an attempt to optimise the perfor-
mode then occurred in 3% of the modules (58), finally mance of the optics and improve the receiver reliability.
spreading into all the arrays. This optimisation work resulted in three new •UCLID•S
The main problems identified from previous concentrators, each of 2 kWp, which have been manufac-
•UCLID•S experience include: tured and installed at three different test sites: Stuttgart in
Germany, Ferrara in Italy and Madrid in Spain. These small
" The maximum concentration level at the focal line is too concentrator systems served as test beds to resolve the tech-
high, producing an unacceptably high flux peak in the nical problems found in the previous •UCLID•S systems.
cells. Once these problems were resolved, answers to the follow-
" The receiver manufacturing process is too complex. ing questions could be provided: Can the •UCLID•S tech-
" Inadequate or insufficient component qualification nology be cost effectively commercialised? Can its
procedures. strengths and weaknesses be clearly identified? To what ex-
" Insufficient or unreliable electrical insulation. tent are these questions and answers applicable to other low
" Sub optimal bypass diode arrangement. to medium concentration systems?
" Mirror reflective surfaces are degraded rapidly by the
weather.
" •xcessively high operating voltage. 2. DESlGN AND MANUFACTURlNG
OF THE NEW EUCLlDES lll
Despite the problems associated with the mirrors and
receivers in the Tenerife power plant, valuable and useful In pursuing the detection and correction of optical defects
knowledge was acquired by the constructing team and by in the •UCLID•S III modules and mirrors and their
Prog. Photovolt: Res. Appl. 2012; 20:356 371 © 2011 John Wiley & Sons, Ltd. 357
DOI: 10.1002/pip
Third generation EUCLIDES concentrator results M. Vivar et al.
combined operation when mounted in the tracking array, it •UCLID•S II Tenerife plant in 1998, this film was not
was assumed that constructing an array made of 16 mirrors available and so several alternative options were deployed:
and 14 modules, providing 20 m2 of collecting area, would
function as a suitable test bed. In order to reduce the optical 1. Mirrors made of polished aluminium of 0.5 mm
mismatch found in previous •UCLID•S systems operating thickness laminated (Metalloxyd, Alanod, •nnepetal,
at 32X and 38X, two smaller gain ratios were explored in Germany) to an aluminium plate 1.5 mm thick.
this project. One side of the array was equipped with a re- 2. Mirrors with silvered film for indoor use (3M) cov-
ceiver operating at 12X, and the other side was equipped ered with a transparent acrylic film and acrylic adhe-
with cells operating at 24X. Because of single axis tracking, sive and laminated together to an aluminium
a pair of extra mirrors was needed to be installed on the plate 1.5 mm thick.
southern end, in the direction of the equator, in order to pro- 3. Mirrors made of polished aluminium plates of
vide continuous illumination on the first receiver over the en- 1.5 mm thickness covered with evaporated silver,
tire year. This is why 16 mirrors, arranged as 8 pairs, are then coated with transparent modified polyethylene
required for 14 receivers, as shown in Figure 1. The overall film on top (Alcoa, Ä„ittsburgh, Ä„ennsylvania, USA).
mirror length along the array is then 8 × 1.174 m = 9.4 m.
An additional metre must be added to accommodate the The metallic mirrors from Alanod had very good optical
structural support and driving elements, making the total performance, but they proved to be very sensitive to the cor-
array length 10.4 m. From these initial specifications, the rosive, saline and humid atmosphere around the acid lands of
key components of the system, mirrors, cells, receivers the volcanic island of Tenerife, particularly near the sea.
and structure, have been reviewed and optimised. Similarly, the plastic metallised thin films (3M and Alcoa)
failed to protect the silver from these harsh conditions.
2.1. Mirror In the new •UCLID•S III, the concentrator mirrors form-
ing the linear parabolic collectors, each with a focal length of
In order to maintain compatibility with previous systems 600 mm, dimensions of 1247 × 1174 mm2 and an aperture
and because the mirror shape was considered optimal in area of 1.464 m2, are made of clear glass 3 mm thick, sil-
the past, the same mirror design from the first generation vered and protected for outdoor operation conditions. Mir-
and second generation •UCLID•S projects was used ror manufacturing was subcontracted to the Spanish
again (Table I). company Induglas. The mirror manufacturing process used
With respect to manufacturing, the mirrors from the thermal slumping, which involved heating and softening
•UCLID•S I prototype incorporated reflective •CÄ„ 305 the glass over an open iron mould, allowing bending into
film from 3M (Maplewood, Minnesota, USA) laminated the desired shape by gravity, then chemically silvering
to aluminium plates shaped to the correct parabolic profile. the back side. Finally, the silver was covered with copper
This film, with 92% reflectivity, was made of acrylic on and protected with polyurethane paints for outdoor use.
which silver was evaporated and back covered to with- Mirror characterisation, conducted at the I•S, included
stand outdoor conditions. However, when developing the measuring the mirror reflectance, which was found to be
Figure 1. Schematic of the EUCLIDES III concentrator set up, showing the mirror deployment, with a total of 16 mirrors, 8 for each
concentration level, and up to 14 receivers, 7 for each row. Two extra mirrors are added to the southern end to provide full illumination
for the first receiver. The tracking system, consisting of a driving wheel and a linear actuator, is placed on the northern end.
358 Prog. Photovolt: Res. Appl. 2012; 20:356 371 © 2011 John Wiley & Sons, Ltd.
DOI: 10.1002/pip
M. Vivar et al. Third generation EUCLIDES concentrator results
TabIe I. EUCLIDES III concentrator mirror design characteristics: dimensional information.
Parameter Symbol Value
Aperture width (perpendicular to axis) [mm] A 1280
Aperture length (parallel to axis) [mm] L 1174
Aperture area [m2] A·L 1502
Height [mm] K 771
Focal length [mm] F 600
f number 0.569
Glass thickness [mm] GT 3
Glass density [kg/cm3] ÁG 2.5
Steel density [kg/cm3] ÁS 7.93
Total mirror weight [kg] 14.75
Mirror profile equation y = x2/4F
87.3% in the wavelength band 350 to 1200 nm, and mea- heavy copper plated grooves. However, the key character-
suring the light flux distribution profile of the focussed istic of interest for •UCLID•S III is that the laser grooved
beam across the receiver. In Figure 2, which shows the ac- front contact pattern is readily modified at low cost for op-
cumulated collected light as a function of the receiver size, timum operation at higher light levels, because the LGBC
it can be observed that about 90% of the light is collected concentrator cells can be manufactured in a high volume
within a width of 4 cm. This means that a solar cell at least production line and they are fully compatible with standard
4 cm wide, which corresponds to a large area CÄ„V solar LGBC manufacturing process. The metallisation process
cell considering a length of 116 mm, will be required. This and the connecting grid design of the one sun LGBC solar
is far from the narrow focal line width of 1.8 cm that was cell from the Saturn line of BÄ„ Solar was modified to work
obtained with laminated aluminium plate mirrors from pre- under concentration levels of 12X and 24X [11]. The main
vious •UCLID•S systems, which produced real operating parameters that were varied include the finger pitch, the
flux distribution close to the design specification. This sub groove depth and the metal layer thickness, which in this
optimal result with glass mirrors motivated the decision to case was copper. The active widths of the two cell types
use larger cells, operating at 12X, to accommodate the mir- were 52.2 mm for the 24X cell and 111.36 mm for the
ror optical characteristics. 12X cell. The length was 116 mm for both cell types. Av-
erage efficiencies of 17.8% were achieved for 12X cells at
1 W/cm2 and 18.5% for 24X cells at an irradiance level of
2.2. Concentrator solar cells
2 W/cm2.
For the third generation •UCLID•S, laser grooved buried
contact cells (LGBC) from the commercial Saturn line for 2.3. Receiver
the flat Ä„V panels of BÄ„ Solar (Madrid, Spain), the same
type of cells used in previous prototypes, were used. Be- One of the key elements in a concentration system is the
cause of the specific characteristics of these LGBC cells design of the receiver module. The receiver houses the
[10], they can be used at low and medium concentration ra- concentrator cells and isolates them from the environment,
tios. Some of these properties include low series resistance, electrically insulates them from the rest of the system, and
low shading losses and the ability to carry high currents in provides the external electrical wiring. Additionally, the
Figure 2. Proportion of energy captured by the concentrator receiver, for a given receiver width centred on the focal line, expressed as
a percentage of total reflected light. Three types of mirrors with different materials and thicknesses are shown.
Prog. Photovolt: Res. Appl. 2012; 20:356 371 © 2011 John Wiley & Sons, Ltd. 359
DOI: 10.1002/pip
Third generation EUCLIDES concentrator results M. Vivar et al.
receiver must extract excess heat and transfer it to the cool- protected from the environment with a glass cover and
ing system, avoiding overheating the cells and receiver clear silicone filler. Finally, two secondary mirrors were
materials. Because the •UCLID•S concentrator receiver also added in order to decrease optical mismatch, by help-
is linear, large area silicon solar cells must be arranged in ing to correct the optical errors introduced mainly by mis-
a single row. When working with single axis linear sys- alignments throughout the long structure, as shown in
tems, the use of multiple single mirrors arranged in a row Figure 3.
produces discontinuities or flux variations along the focal The receiver manufacturing process was conducted at
line because of the gaps between individual mirrors, so the I•S in a fully manual process. Forty three receivers
long cells must be used in order to minimise this optical were made, 21 for the 24X concentration level and 22 for
mismatch across any one cell. The mounting, heat sinking the 12X system. They were characterised using a flash so-
and encapsulation of large area solar cells is a difficult task lar simulator plus I V equipment, returning average effi-
because of differential thermal stresses caused by dissimi- ciencies of 17.4% for the 24X receiver and 16.6% for the
lar materials, thermal processes used during cell encapsula- 12X receiver under uniform illumination of 2 W/cm2 and
tion, and thermal excursions encountered during real 1 W/cm2, respectively, at 25°C. Series resistance was also
operating conditions. At the same time, large area cells re- measured by acquiring the dark I V for each receiver.
quire less handling steps per unit area, which is a signifi- The results show a very important reduction in series resis-
cant factor in assembly cost reduction. tance, with respect to previous projects with the same cell
In general terms, a critical aspect [12] of concentrator geometry and size, from 10 m© for the •UCLID•S II to
receiver design involves the selection of the interface mate- 5.9 m© (average) for the •UCLID•S III receiver with
rial between the cells and the heat sink that ensures good 24X cells.
electrical insulation, good thermal performance, good ad-
hesion properties and tolerance to the repeated mechanical 2.4. Supporting structure
stress introduced by thermal cycling between materials
with differing thermal properties. The last point is impor- Because of the small collecting area of the •UCLID•S III
tant in concentrators, because the effect of clouds can sud- concentrator, a variation involving a small addition to the
denly reduce high power density on the cell to zero and length of the previous •UCLID•S II structure was con-
then back to high again. Small cells are less affected by this structed. This allowed the driving wheel to be located at
mechanical stress which produces fatigue in the interface the north end, close to the north support and outside the
materials and surfaces. The •UCLID•S I II receiver de- mirror area. The sites chosen for systems deployment and
sign used a thermal film as a cell to heat sink interface ma- test were Ferrara, Stuttgart and Madrid. For the Madrid
terial for glueing the cell to the heat sink, thus ensuring site, the old •UCLID•S I structure was adapted, whereas
thermal conduction and electrical insulation. However, this for Ferrara and Stuttgart, the new small systems were made
material failed in the Tenerife system after a short period of by the Spanish company Jupasa at their factory and sent
time, because at high temperatures the thermal film adhe- abroad, as shown in Figure 4.
sive properties diminished, and consequently the heat
transport characteristics declined. Moreover, this solution
demonstrated poor performance as an electrical insulator, 3. QUALlFlCATlON TESTlNG
producing shorts from the cells to ground and subsequently
the failure of the entire system. The set of failures that occurred with the •UCLID•S II
In view of these considerations, for •UCLID•S III sys- Tenerife power plant provides a clear example of why
tem design, alternatives to thermal film such as direct cop- deep and specific qualification and reliability testing of
per bonded substrates (copper alumina copper), insulated system components are critically important. What was
metal substrates (copper polyamide metal) and alumina initially planned as the largest demonstration CÄ„V plant
plates were analysed and tested before selecting the opti- in the world at that time, as a new challenge in the de-
mum. Results from tests show that direct copper bonded velopment of the concentrator technology, ended up in-
substrates are still too expensive for low concentration sys- advertently as a demonstration of how sensitive the
tems and that insulated metal substrates produces signifi- entire system operation is to specific failures of key
cant bow on the cells after soldering. The third option, components when working under real operating condi-
alumina substrates, was used as an interface material be- tions. Although  in house qualification and type ap-
tween the cells and the heat sink, glueing all the materials proval tests were performed before the fabrication of
together with a thermal epoxy. This low cost solution pro- modules and mirrors, among other components, were
vided good thermal transfer so it was selected as the inter- commenced, these two items failed in the field. Some
face material for the new receiver. of the mirrors (15% of the total number) degraded rap-
Finally, the complete receiver design consists of an alu- idly when exposed to outdoor conditions, and the
minium base electrically isolated by alumina plates on the receivers also failed, exhibiting electrical insulation
top, which the solar cells were glued with a thermal epoxy. defects and heat transfer problems.
•ach solar cell had its own individual bypass diode, which The component failures at the Tenerife plant were ana-
was fully integrated between the cell tabs. The cells were lysed, especially the receiver modules, in order to identify
360 Prog. Photovolt: Res. Appl. 2012; 20:356 371 © 2011 John Wiley & Sons, Ltd.
DOI: 10.1002/pip
M. Vivar et al. Third generation EUCLIDES concentrator results
Figure 3. Image of the final EUCLIDES III receiver, including the cells, the heat sink and the secondary mirrors.
the principal causes of failure to prevent the most common new specific standard, the draft I•C 62108 intended for
problems that could be found in similar concentrator sys- CÄ„V systems, was close to approval. Based on this standard,
tems. A range of different conclusions were drawn, but a different set of tests were conducted for the two main inno-
the main theme was the lack of reliability of the most inno- vative component designs: mirrors and receivers. However,
vative elements. It is important to note that, at the time, no no laboratory was available and accredited to perform the
international standards were available for testing concen- complete test plan required in the I•C 62108 standard.
trating Ä„V components. It is obvious that the methods
and tests proposed for the prototypes were insufficient
3.1. Mirror qualification
and inadequate, as they did not reveal the short term fail-
ures of the components subsequently displayed in the field.
The new second surface •UCLID•S mirrors were made of
The impact of the problems of the Tenerife plant, which
3 mm clear glass, with silver and copper layers on the rear
was well planned in many other respects, caught the attention
surface, protected by a coating of paint. In order to test the
of the CÄ„V community and dramatically, if not intentionally,
sealing properties of this protective layer, salt mist tests
revealed the need for establishing a qualification standard spe- were performed according to the I•C 60068 2 11. The
cific for CÄ„V, because the reliability tests designed for flat
mirror sample passed the salt mist test without degradation
plate systems were shown to be unsuitable for CÄ„V. When
once the edge sealing by the paint coating was assured.
developing the third generation •UCLID•S concentrator, a
Another test carried out on the mirrors was the hail impact
Figure 4. New EUCLIDES III structure manufactured as a smaller version for the prototypes of Stuttgart and Ferrara.
Prog. Photovolt: Res. Appl. 2012; 20:356 371 © 2011 John Wiley & Sons, Ltd. 361
DOI: 10.1002/pip
Third generation EUCLIDES concentrator results M. Vivar et al.
test, conducted at the Centro de Investigaciones •nergét- In this case, the standard was especially useful for
icas, Medioambientales y Tecnológicas, Madrid facilities. detecting deficiencies in the receiver design. In the first
The collector was subjected to normal impacts using test, the module substrate became separated from the heat
25.4 mm diameter ice balls at a speed more than 23.1 m/s, sink to which it was bonded. Following a failure analysis,
according to the I•C 62108. The mirror passed the hail test it was determined that the epoxy adhesive, which had been
without any damage. Finally, the adhesion performance of filled with alumina thermally conductive material, used to
the mounting supports was tested for several back cover glue the modules to the heat sinks was in a poor condition.
paints. This was necessary because the mirror fixtures were Following the standard procedures, a second sample was
bonded onto the protective paint surface. Because the I•C manufactured using a fresh adhesive. After the subsequent
62108 standard does not specify tests for this qualification thermal cycling test, the electrical performance of the sam-
aspect, the mirror mounts were tested according to the ple returned a power degradation of 23.5%, which was
Standard TA 301/95 09 by the Sika Company in Madrid. well above the acceptable limit of 8%. Therefore, although
The mounting specification passed the test when outdoor the thermal bonding survived, the sample did not pass the
polyurethane paint was used as a sealant. test and the receiver design should be reviewed.
These I V curves of the module before and after the
degradation due to thermal cycling test, shown in Figure 5,
3.2. Receiver qualification , indicate a photocurrent reduction and an increase in series re-
sistance. The solution for avoiding this degradation was not
The procedures detailed in draft K of the I•C 62108 stan- identified immediately, but the contract schedule required
dard were followed for the •UCLID•S III receiver qualifi- progress on the manufacturing of the receivers. Special atten-
cation. The equipment used to perform the tests was tion was given to electrical insulation, thermal performance
located in several UÄ„M laboratories. Because of the re- and procedures for soldering tabs to the cells [8]. It was not
duced dimensions of the available climatic chambers possible from these early results obtained using the I•C
(50 cm × 50 cm × 50 cm), four representative samples of 62108 tests to determine whether the receivers were im-
24X receivers, which were smaller than the real size recei- properly manufactured for the intended application or
vers, were manufactured, including all the constituent com- whether the qualification standard was too demanding.
ponents. The assembly procedures followed the same
manufacturing and characterisation processes as specified
for the full size receiver. The first test procedures consisted 4. EUCLlDES lll SYSTEM
of visual inspection, electrical performance with natural or lNSTALLATlON AND FlELD
simulated sunlight, electrical insulation test and wet electri- PERFORMANCE
cal insulation test, according to the methodology specified
in the I•C 62108 standard, were common for all the sam- After the components were manufactured, three systems,
ples. To continue with the test, four different programme each of 24 m2 collecting area, were assembled and in-
sequences were constructed: stalled, as shown in Figure 6, in three different test sites:
Stuttgart (Germany), Ferrara (Italy) and Madrid (Spain).
" Sequence A: Bypass diode thermal test (with the sample •ach system comprised two rows of eight mirrors and
at 75°C, apply ISC through the receiver for 1 h, then two rows of seven series connected receivers, one row
measure bypass/blocking diode temperature, and then corresponding to the 24X concentration level on the east-
apply 1.25 * ISC for additional 1 h.) ern side and the other row to the 12X concentration level
" Sequence B: Ä„re thermal cycling and humidity freeze on the western side. In principle, no systematic errors were
test (85°C and 85% relative humidity for 20 h followed anticipated because of this configuration of two rows of
by 4 h cool down to -40°C, 20 cycles). mirrors, except for the possibility of differing weight load
" Sequence C: Thermal cycling (500 cycles from -40 to distributions during the course of a day for the mirrors,
110°C, apply electrical cycles with a cycle speed of 10 and only then if the mirrors support structures were not
electrical per each thermal when T > 25°C and inject a sufficiently rigid. The installation steps of a system com-
current of 1.25 * ISC). prising 24X and 12X concentrators were as follows: struc-
" Sequence D: Damp heat test (1000 h at 85°C and 85% tural installation, followed by receiver mounting and
relative humidity) and water rain test (1 h water spray electrical connection, mirror installation and trimming
on each of four specified orientations). and finally installation of the tracking control system.
The structures for the Ferrara and Stuttgart sites were
Two objectives were pursued with these tests: firstly, to shipped by truck as a disassembled unit, requiring only a
validate the receiver design, and secondly, to test the stan- small crane for assembly but without requiring any weld-
dard itself, as this was the first team [13] as far as is known, ing in the field. It was necessary to add a temporary load
to check whether the tests really provided reliable and ef- on the driving bar end to maintain balance during the fixing
fective means of detecting operating failures. After con- process.
ducting this set of test sequences, the receiver passed all During set up, the receivers were in a fixed position
except for the thermal cycling test. whereas the mirrors were mobile, allowing adjusting of
362 Prog. Photovolt: Res. Appl. 2012; 20:356 371 © 2011 John Wiley & Sons, Ltd.
DOI: 10.1002/pip
M. Vivar et al. Third generation EUCLIDES concentrator results
4.1. Electrical performance
The electrical performance of the installed systems was
monitored in order to compare the different systems
with the two concentration levels, 12X and 24X, as
well as to compare the system performance in locations
with different climatic conditions. The main performance
characteristics from manufacturing and early testing of
the receivers and mirrors are shown in Table II. There
were two groups of available data, corresponding to the
Madrid and Stuttgart systems, for each of the two levels
of concentration (24X and 12X), over a span of three
months (because of unforeseen project reasons, no data
were available from the Ferrara test site). For each system,
instantaneous I V curves, electrical mismatch, optical mis-
Figure 5. Results from thermal cycling tests of the EUCLIDES
match, accumulated production and total losses were
III receivers (-40°C to 110°C plus electrical cycles). I V curves
measured.
from before and after the test show the degradation of the
The results of the analysis of one of the systems will be
receiver.
presented, followed by a comparison between all the in-
stalled concentrators.
A study during February 2007 of the electrical perfor-
the focal line on the cells by means of trimming screws. mance of the •UCLID•S III 12X, installed in Madrid,
Trimming was conducted by measuring the short circuit shown in Figure 7, illustrates the maximum output power
of the receivers. Once the mirror positions were trimmed, as a function of effective irradiance onto the system during
each row of receivers was connected in series. a typical day in February. The systems have a tracking ex-
From these three systems, performance data is only cursion of Ä…70° with respect to zenith, which causes the
available for the Madrid and Stuttgart concentrators. For abrupt drops near sunrise and sunset. The curve follows
each system and concentration level, the electrical and reasonably the direct irradiance, corrected by the cosine
thermal performance has been obtained and analysed. Sys- of the angle of the sun with the normal to the collector ap-
tem production has been modelled and compared with real erture, quite closely. Figure 8 shows several I V curves
operational data, and the effect of soiling [14] on the sys- recorded during the course of the day. The effect of optical
tems has been studied. This information was used to ap- mismatch between the receivers can be seen very clearly
proximate the maximum allowed system cost for because the system consists of seven modules in series.
commercial competitiveness. With only seven modules, the effect of the optical
Figure 6. EUCLIDES III concentrator installed at Stuttgart (Germany), including the two arrays for the 24X and 12X concentration
levels.
Prog. Photovolt: Res. Appl. 2012; 20:356 371 © 2011 John Wiley & Sons, Ltd. 363
DOI: 10.1002/pip
Third generation EUCLIDES concentrator results M. Vivar et al.
mismatch is more visible, because individual bypass the worst receiver modules because of the small produc-
diodes act at different times, clearly showing the steps in tion run.
the I V curves. In a full system with, for example 200 To complete the electrical performance discussion,
modules, the effect would be smoother in the total I V the accumulated production of the two systems, shown
curve of the system because the number of bypass diodes in Figure 10, for their two concentration ratios, was ana-
increases, as well as the final value of system voltage. lysed for the 3 month duration of this study. The
What can be seen clearly in this figure is the variation of •UCLID•S 12X concentrators achieved higher production
the I V curve as a function of mirrors profile changes during the winter months because of the lower position of
caused by the array orientation. This indicates that the mir- the sun that produces higher cosine losses, as well as the
rors are not sufficiently rigid because they have no ribs and fact that the •UCLID•S 12X system has a larger receiver
are also very heavy and unbalanced, so the focal profile area and therefore more tolerance than the •UCLID•S
deforms differently during the day because of tracking ori- 24X system. However, during the rest of the year, the
entation. To avoid this situation, it would be necessary to •UCLID•S 24X had higher production, principally be-
have lighter mirrors and more rigid or simply a greater cause the receiver efficiencies were almost one percentage
number of support structures. point higher, as described in Section 2.3.
Finally, the output system power versus the input power,
which in this case is the effective irradiance, cosine corrected,
4.2. Thermal performance
is presented in Figure 9. This record shows great variation
with respect to the ideal linearity, primarily caused by elec-
In order to evaluate the thermal performance of the system
trical and optical mismatch between modules, with the lat-
and then to convert the measurements from real operation
ter optical mismatch being the most important. The
conditions to standard conditions (STC), two methods
optically based losses arise from the discontinuities be-
can be used to determine the cell temperature:
tween the mirrors and misorientation of mirrors but primar-
ily from the previously mentioned deformation, which is
equivalent to the defocussing of the concentrated beam. 1. Use of the specific thermal resistance Rth measured
Combining the results from the Stuttgart •UCLID•S III between the cell to the ambient air or to any other
prototypes with the Madrid prototypes, calculated system part of the receiver. By measuring the reference tem-
efficiencies were in the range of 10% to 12%, as shown perature and then using the Rth, the cell temperature
in Table III. The drop in efficiency, from the original 17 is calculated.
18% cell efficiency, is quite significant and requires the 2. Use of the cell temperature coefficient ²(C), which
identification and measurement of the factors contributing measures the VOC variation as a function of the tem-
to the output power loss. Optical factors such as mirror perature T and the concentration level C. By know-
reflectivity, the gap between mirrors, non uniform illumi- ing this coefficient and the concentration level and
nation distribution on cells, geometric errors of the mirror the VOC of the system, the cell temperature can be
profile, and variable mirror deformation during the course
easily obtained.
of the day are calculated to a loss of 16% of the light effec-
tiveness for electricity generation in the concentrator re- For the •UCLID•S III modules, the value of Rth was
ceivers. The electrical mismatch is higher than usual,
determined by injecting electrical power to a small sample
primarily because of the reduced number of cells and
module and studying the thermal management. The Rth
modules built for this project, which limited cell and re- was measured, obtaining a value of 20.5°C cm2/W be-
ceiver matching opportunities, with a significant variation
tween the cell and the ambient air. This means that at
of cell characteristics and final manufactured module
12X, the difference between the cell temperature and the am-
parameters. Unfortunately, it was not possible to reject
bient air temperature is approximately 40°C. Therefore, in the
TabIe II. EUCLIDES III main concentrator parameters: 24X and 12X systems.
EUCLIDES III system parameters 24X 12X
Module interconnection Series Series
Number of modules 77
Number of mirrors 88
Collector net aperture area [cm2] 105 140 105 140
Active receiving area [cm2] 3654 7795
Geometric concentration [suns] 25.58 11.80
Number of cells 70 70
Thermal resistance [°C cm2/W] 20.5 20.5
Series resistance including wires [©] 0.122 0.126
364 Prog. Photovolt: Res. Appl. 2012; 20:356 371 © 2011 John Wiley & Sons, Ltd.
DOI: 10.1002/pip
M. Vivar et al. Third generation EUCLIDES concentrator results
Figure 7. EUCLIDES III Madrid electrical performance: maximum power point (MPP) versus effective irradiance (Beff) during the day
(data corresponding to 1 February 2007).
case of the Madrid system, operating in autumn and spring Figure 11, for example, one receiver module with signs
seasons when temperatures are about 25°C, the cells are oper- of bad heat transfer can be seen under operation, with its
ating at 70°C to 80°C, in summer this can be up to 80 90°C, associated infrared image, revealing temperatures up to
decreasing in winter to 50 60°C during sunny and clear days. 130°C.
In addition, the cells were also thermally characterised, so cell
temperature coefficients were also available [11].
4.3. Modelling
Under real operation conditions, some modules failed be-
cause of thermal adhesive failures, with cells reaching tem- System field performance can be modelled in order to pre-
peratures up to 100°C, whereas other receiver modules
dict the annual energy production in a specific location.
were operating properly at temperatures of 70°C. Infrared
I•S UÄ„M developed a system model with the objective
imaging was used to detect the receivers with bad heat trans- of trying to predict the annual energy production of differ-
fer in early stages. By using this technique, potentially faulty
ent concentration systems using silicon solar cells [15].
modules could be removed and analysed early while keeping
This model was based on an I V curve from the system,
the rest of the system keeping safe working conditions. In
along with the corresponding meteorological conditions
Figure 8. EUCLIDES III Madrid electrical performance. Different I V curves during the day show the optical mismatch produced by the
mirror deformation and the effect this has on the electrical performance (data corresponding to 1 February 2007).
Prog. Photovolt: Res. Appl. 2012; 20:356 371 © 2011 John Wiley & Sons, Ltd. 365
DOI: 10.1002/pip
Third generation EUCLIDES concentrator results M. Vivar et al.
Figure 9. EUCLIDES III Madrid electrical performance: maximum power point (MPP) versus effective irradiance (Beff) shows the data
dispersion due to losses in the system (data corresponding to 1 February 2007).
of the site. The forecast modelling uses the I V curve cor- the focus to the average irradiance), m is the diode quality
rections shown in •quations 1 and 2 to calculate the values actor, k is the Boltzmann constant (1.38 × 10-23 J/K), q is
of I and V at other irradiance conditions, which may be dif- the elementary charge of an electron (1.6 x 10-19), Å‚ is
ferent from the real operating conditions. Using this the index, which fluctuates between 2 and 4, ISC1 is the
method, the model calculates the energy production over measured short circuit current and ISC2 is the short circuit
the course of the year, integrating the different climatic current at the selected conditions.
conditions for each day. This model was programmed The •UCLID•S III systems, at their two different con-
within the framework of the •uropean Ä„roject C Rating centration levels of 24X and 12X, were modelled [16] in
in a software tool named  Concentrator Modeling by the order to obtain the annual energy production, shown in
research team and can be used to predict the annual, Figure 12, for the three different climatic conditions
monthly or daily energy production. corresponding to the ID•OCONT• project test sites: Fer-
rara (Italy), Stuttgart (Germany) and Madrid (Spain). In or-
der to compare the •UCLID•S III concentrator with
B2
I2 ź I1 þ ISC1 -1 þ Ä…ðT2-T1Þ (1)
conventional flat panel technology, data from the I•S
B1
building integrated Ä„V generator was used to model its an-
nual energy production. Two cases were analysed: a static
where I1 and V1 are the coordinates of a point of the mea-
Ä„V system, and a two axis tracking system. The results for
sured characteristics, ISC1 is the measured short circuit cur-
the •UCLID•S 12X and •UCLID•S 24X systems pro-
rent, T1 is the measured cell temperature, T2 is the cell
vided an approximately similar yield for all three locations.
temperature at the selected ambient temperature, B1 is the
However, it can be seen that the •UCLID•S 12X would
measured irradiance incident on the collector, B2 is the se-
have a slightly higher energy production than 24X in the
lected irradiance incident on the collector and Ä… is the cur-
Stuttgart and Ferrara test sites. Considering the fact that
rent temperature coefficient of the cell.
the 12X receivers have lower efficiencies than the 24X
ones by almost one percentage point, it would appear that
EG T2
V2-V1 ź -V1 1- (2)
the 12X system would perform better in this type of cli-
q T1
mate. On the other hand, the model indicates that the
mkT2 Isc2 T2
•UCLID•S concentrators would have similar performance
þ ln -Å‚ ln -½I2-I1ŠRarrayÅ"U
q Isc1 T1
to static flat panel but that the performance would remain
lower than a two axis tracking flat panel system.
where I1 and V1 are coordinates of points of the measured When comparing the real energy production with the
characteristics, I2 and V2 are coordinates of the predicted energy production, the result showed that real en-
corresponding points on the corrected characteristics, Rarray ergy production was lower than predicted energy, as
is the total array series resistance, T1 is the measured cell shown in Table IV. There are several reasons for this, but
temperature, T2 is the cell temperature at the selected the main cause is that the software works with a certain
ambient temperature, EG is the cell band gap, U is the I V curve, without considering all the factors that affect
non uniformity factor (ratio of the maximum irradiance at the real energy production, such as the electrical and
366 Prog. Photovolt: Res. Appl. 2012; 20:356 371 © 2011 John Wiley & Sons, Ltd.
DOI: 10.1002/pip
M. Vivar et al. Third generation EUCLIDES concentrator results
TabIe III. Comparison of results from the instantaneous electrical field performance of the EUCLIDES III concentrators at standard
conditions (1000 W/m2, 25°C).
Parameter EUCLIDES Madrid 24X EUCLIDES Madrid 12X EUCLIDES Stuttgart 24X EUCLIDES Stuttgart 12X
Open circuit voltage [V] 46 42 49 44
Short circuit current [A] 27 33 25 27
Maximum power point [W] 960 835 967 990
System efficiency [%] 11 10 11 12
optical mismatch of the system, which as previously docu- concentration systems, a point focus refractive concentra-
mented, varies over the course of the day because of mirror tor and a flat panel module. The concentration systems
deformations. Other reasons for larger differences between used were the Archimedes 2X and Archimedes 10X from
predicted energy and real energy production are the peri- ZSW (Germany), the •UCLID•S 12X and •UCLID•S
ods on which the systems are not operating because of 40X from the I•S UÄ„M, and a point focus Fresnel lens
maintenance operations or delays in the setting up of the based system operating at 300X. All the concentrators
plant (for example, in January the predicted energy is 10 were installed at the I•S UÄ„M test facilities in Madrid,
times higher than actual). Spain.
4.4. Soiling
The effect of soiling in CÄ„V systems was also analysed
during this project [15,17]. Some experiments were con-
ducted at the I•S UÄ„M sites, including linear reflective
a
b
Figure 10. (a) EUCLIDES III Madrid production for the two con-
centration levels for a testing time period of 3 months. (b) Figure 11. (a) Visual image of a EUCLIDES III module under op-
EUCLIDES III Stuttgart production for the two concentration eration. (b) Thermal image of the same faulty module with bad
levels for a testing period of 4 months. heat transfer and temperatures up to 130°C.
Prog. Photovolt: Res. Appl. 2012; 20:356 371 © 2011 John Wiley & Sons, Ltd. 367
DOI: 10.1002/pip
Third generation EUCLIDES concentrator results M. Vivar et al.
of two interactions with the particles of dirt and so the
losses can be more substantial. On the other hand, systems
using reflective optics close to one surface, as the Archi-
medes systems, which have a very thin mirror, suffer less
loss because of soiling than concentrators with a second
surface optics, such as the •UCLID•S optics which have
a 3 mm thick glass in front of the silver layer. Second
surface optics suffers higher reductions in efficiency be-
cause of soiling, for the same reasons detailed previously.
It is important to take into account the effect of soiling
in CÄ„V systems because the reduction in the energy output
in certain cases can be quite significant. However, accu-
Figure 12. Comparison of the predicted yield for the EUCLIDES
rately quantifying losses due to soiling is very difficult.
III concentrator (24X and 12X) and for a flat panel static system
The problem should be studied in relation to each system,
and two axis tracking system, showing the higher predicted per-
while bearing in mind that the specific characteristics of the
formance of the two axis tracking flat panel system versus the
particular location may dictate a particular cleaning
EUCLIDES III concentrator, with similar performance to a static
regime.
flat panel system.
5. COST ANALYSlS
The Archimedes 2X is a CÄ„V system with  V mirrors
that concentrate the sunlight to about 2 suns onto the solar
The maximum allowed cost of the •UCLID•S III technol-
cells, which are passively cooled with fins on the back of
ogy [8] for the system to be competitive in the current mar-
the modules. Archimedes 10X is a cylindrical parabolic
trough with a concentration level of 10X. Solar cells are lo- ket, taking into account the shortage of silicon feedstock
and the current feed in tariffs of some countries, such as
cated in the back of the mirrors, above which is placed the
passive finned cooling system. The Fresnel 300X concen- Italy and Spain, has been analysed. The main objective is
to achieve a •UCLID•S system cost that can be fully re-
trator module is a point focus system using lenses and
covered in a period of 10 years under the current Spanish
equipped with two axis tracking. The flat monocrystalline
feed in tariff conditions. The feed in tariff for Ä„V systems
module was from BÄ„ Solar.
All the systems were measured when soiled and then af- in Spain has been published in the  Real Decreto 436/
2004 of 12 March 2004. The relevant price for the energy
ter cleaning, which produced proportionately different
produced in installations with an instantaneous power up to
increases in ISC for each system, as shown in Table V. In
100 kWp was Ź 0.440381/kWh during 2007 [18].
general, the results showed that CÄ„V systems are more
sensitive to soiling than flat panels, accumulating losses
in ISC of about 14% on average for the three different tests
5.1. Energy production calculation and cost
conducted at I•S UÄ„M. In the case of the •UCLID•S con-
per watt
centrators, losses can reach up to 26% when the system is
soiled over 4 months of continuous exposure.
In order to calculate the maximum allowable cost for the
In general, it would be expected that linear reflective
system, it is necessary to analyse three variables: firstly,
optics would suffer greater losses because of soiling than
the electrical energy production capability of the system;
refractive optics. In a refractive optical device, the light
secondly, the income from the feed in tariff, including bo-
has only a single passage through the soiled layer. How- nus and incentives; and finally, the standard test conditions
ever, for a reflective surface, the light has the possibility
to obtain the cost per watt peak. We have used the
TabIe IV. Modelling of the energy production of the EUCLIDES III systems for Stuttgart and comparison with real energy production.
Energy production [kWh]
Madrid Stuttgart
24X 12X 24X Predicted 24X 12X Predicted 12X
January   2.3 23.0 2.5 25.0
February 18.8 21.3 10.4 55.0 10.7 53.0
March 39.8 30.3 56.0 84.0 53.8 77.0
April 62.9 36.1 101.8 120.0 77.3 106.0
May   48.6 133.0 0.5 114.0
June       
368 Prog. Photovolt: Res. Appl. 2012; 20:356 371 © 2011 John Wiley & Sons, Ltd.
DOI: 10.1002/pip
M. Vivar et al. Third generation EUCLIDES concentrator results
TabIe V. Results from the soiling study in concentration photovoltaic systems versus flat panel systems over a period of approxi-
mately 3 4 months. It can be observed that the concentrator systems are more sensitive to soiling than flat panels.
System ISC before cleaning [A] ISC after cleaning [A] Losses due to soiling [%]
November 2007
Archimedes 2X 22.2 23.7 6.5
Archimedes 10X 15.1 18.1 16.8
Flat panel BP 2.5 2.6 0
March 2008
Flat panel BP 2.9 2.9 1.2
Archimedes 2X 21.5 23.3 8
Archimedes 10X 10.7 12.7 15.5
Euclides 24X 22.1 30 26.2
Euclides 12X 9.7 12.6 23.2
Fresnel Module 300X 3.9 4.5 12.3
July 2008
Flat panel BP 2.7 2.7 0
Archimedes 2X 23.8 24.7 3.9
Euclides 24X 24.9 31.6 21.2
Fresnel Module 300X 6.9 7.3 5.6
meteorological data from Madrid to calculate the energy been estimated and updated from those obtained from the
produced by a •UCLID•S III 24X system: experience of the •UCLID•S III system and the
•UCLID•S II power plant. The cost of the cells is based
h i
Annual Energy½kWhŠ ź Ba kWh=m2 Å"Ar m2 Å"·optÅ"PRÅ"·rx on the available LGBC technology from BÄ„ Solar during
the development of the three generations of •UCLID•S
ź 1640Å"ArÅ"0:8Å"0:7Å"0:174
systems; with an average cell efficiency of 18%. Table VI
ź 159:7Å"Ar½kWhŠ
shows the expected costs of the •UCLID•S III over
(3)
the short and long term, initially for a production of
2 MW per year, and subsequently for a long term pro-
where Ba is the radiation collected by one concentrator duction of 50 MW per year. Figure 13 presents the total
with one horizontal axis orientated north south in Madrid distribution of component cost, where the cell cost and
(Ba = 1640 kWh/m2); Ar is the collector surface area; ·opt availability are for the moment the main limitation of the
is the optical efficiency (·opt = 80); PR is the  performance system.
ratio (0.7); and ·rx is the receiver efficiency at nominal op-
erating conditions with a cell temperature of 25°C
(·rx = 17.4%).
6. CONCLUSlONS
The net income from the energy produced according to
the 2007 feed in tariff, including taxes and maintenance
With regard to the different objectives established for the
expenses amounting to 20% of revenue per year, is
third generation of •UCLID•S concentrator, the first task
was to determine whether the •UCLID•S technology
Net Income½Å¹ Š ź 159:7Å"Ar½kWhŠÅ"0:440381½Å¹ =kWhŠÅ"0:8
could be successfully commercialised under the current
ź 56:26Å"Ar½Å¹ Š
market conditions. Results from the third generation
•UCLID•S system established a 10% average system effi-
For an array of 250 m2 aperture area, with a peak power ciency, which is similar to the efficiencies of previous
of 33.1 kWp at STC (irradiance = 1000 W/m2, Tcell = 25° •UCLID•S systems, primarily as a result of using the
C), the cost per Wp should be less than Ź 4.3/Wp for a pay- same type of solar cell with the same efficiency. Series re-
back period of 10 years. sistance in later modules improved relative to previous ver-
For the current Spanish Ä„V market conditions prevailing sions of the concentrator, but the final total system losses
in 2010 (Real Decreto 1578/2008), with a reduced were similar.
feed in tariff of Ź 0.32/kWh, the maximum allowable The most significant benefit of this research project was
cost of the •UCLID•S would be only Ź 3.00/Wp. the valuable experience obtained in the system installation,
system monitoring and the analysis of the results. The main
5.2. Calculation of component and system technical improvements for this third generation of
cost •UCLID•S concentrators relate to the industrialisation of
the single components and the improved component and
In order to calculate the total system cost, the cost of the system reliability. Manufacturing processes were opti-
single components was estimated separately. Costs have mised, and some of the components such as the support
Prog. Photovolt: Res. Appl. 2012; 20:356 371 © 2011 John Wiley & Sons, Ltd. 369
DOI: 10.1002/pip
Third generation EUCLIDES concentrator results M. Vivar et al.
TabIe VI. Projected cost of a 33.1 kWp EUCLIDES III 24X array, at standard conditions (STC), for an initial annual production of 2 MW
and a long term production of 50 MW/year. The estimate was based on the cost of materials, labour and installation obtained from the
experience of EUCLIDES III system and EUCLIDES II power plant.
Component Size Unit price Total cost for a 33.1 kWp Total cost for a 33.1 kWp
array with 2 MW of initial array with 50 MW of long
annual production term annual production
[Ź ] [Ź ]
Optics Mirrors 250 m2 Ź 70.00/m2 17 500 17 500
Receiver manufacturing 13 500 9500
excluding cells
Heat sinks 33 250 W Ź 0.40/W 13 300 11 600
Structure, tracking system and 26 000 24 000
tracking control unit
Field preparation and installation 12 000 11 000
Transport 400 km, ground 12 000 11 000
Inverter and grid connection 26 000 Wp Ź 0.30/Wp 7800 7800
Total excluding cells 102 100 92 400
Cells 1730 Ź 10.90 11.20/cell 19 400 18 800
Total system cost 121 500 111 200
Cost per Wp (STC) 3.7 3.4
Figure 13. Estimated capital cost distribution for the EUCLIDES III concentrator, showing the different component contributions to the
final cost. Cost was estimated for a concentrator size of 250 m2 of aperture area and an annual production of 50 MW.
structure, heat sink, tracking system and control system are system rigidity if glass mirrors are used, in order to avoid
now approaching commercial ready products. mirror deformation losses. Considering the previous lami-
The most innovative system elements have also made nated mirrors, it could be possible to use new reflective
significant progress towards reliable, industrialised pro- films with proven outdoors reliability, which are now
ducts. For example, commercial 1 sun LGBC cells, which available in the market.
at the time of the project were manufactured by BÄ„ Solar The new receiver designs guaranteed sufficient electri-
but are now only produced by Narec in the UK, were used cal insulation to the cells provided by alumina plates,
for the project with minimal modifications for operation which also provided good thermal conduction. However,
under concentration. These cells could be used in general large area cells are difficult to encapsulate, and the require-
for low and medium concentration levels, with low series ment for automated processes along with associated costs
resistance, but with an efficiency limit of about 18%. It must be taken into account.
would be interesting to be able to use low concentration In summary, given that the expected cost for the
silicon solar cells with efficiencies up to 23%, manufac- •UCLID•S III system for an annual production of 2 MW
tured at low cost for this concentrator, but they are not is Ź 3.70/Wp, and that the maximum projected allowable cost
yet available on the current market. for the system is Ź 3.00/Wp in the current ĄV market with the
Glass mirrors manufactured for the •UCLID•S III sys- Spanish feed in tariff of Ź 0.32/kWh, it is very difficult to
tems showed poor quality; and despite proven reliability, achieve the required cost targets with the current system effi-
the optics were poor in comparison with the previous lam- ciency. It would be necessary to either increase the total sys-
inated mirrors. Mirror manufacturing should be improved tem efficiency, or reduce the single component costs, or
in order to provide better optical quality and sufficient some blend of both approaches, in order to be able to reach
370 Prog. Photovolt: Res. Appl. 2012; 20:356 371 © 2011 John Wiley & Sons, Ltd.
DOI: 10.1002/pip
M. Vivar et al. Third generation EUCLIDES concentrator results
full commercialisation in the near future. For example, sim- 7. Cunningham DW, Gasson MÄ„, Bruton TM. Develop-
ply by using 23% silicon solar cells, the total system cost
ment of a solar cell module capable of operating under
could be reduced to Ź 2.85/Wp instead of Ź 3.70/Wp.
30x concentration. 13th European Photovoltaic Confer-
ence and Exhibition, Nice, France, 1995; 2278 2280.
8. Vivar M, Optimisation of the •UCLID•S Concentra-
ACKNOWLEDGEMENTS
tor, PhD thesis, Instituto de •nergía Solar, Universidad
Ä„olitécnica de Madrid, Spain, 2009.
This work has been supported by the •uropean Commis-
sion,  •nergy, •nvironment and Sustainable Develop- 9. Sala G, Ä„achón D, Anton I, Vivar M, Mohring HD,
ment , Fifth Framework Ä„rogramme, through the funding Morilla C, Fernandez JM, Martinelli G, Stefancich
of the ID•OCONT• project (Ref: •NK5 CT 2002
M, Malagu C, •ames Ä„C, Mallick TK, Luque I.
00617). It has also been supported by the Comunidad de
ID•OCONT• project: searching the best Si cells Ä„V con-
Madrid within the NUMANCIA Ä„rogramme (S 05050/
centrator. 20th European Photovoltaic Solar Energy
•N•/0310), and by the Spanish Ministry of •ducation
Conference and Exhibition, Barcelona, Spain, 2005.
and Science under the Consolider Ingenio 2010 Ä„rogram,
10. Morilla C, Russell R, Fernández JM, Vivar M, Sala G.
through the Ä„roject G•N•SIS FV (CSD20006 0004).
Technology improvements in buried contact cells opti-
The authors would also like to thank Jose Ąińero, Juan
mised for concentration systems. 4th International
Carlos Zamorano, Angel Barreno and María Martínez from
Conference on Solar Concentrators for the Generation
I•S; Christoph Gruel, Fritz Klotz, Hans Dieter Mohring and
of Electricity or Hydrogen, •l •scorial, Spain, 2007.
Janine Jäck from ZSW; and Giuliano Martinelli and Marco
11. Vivar M, Morilla C, Anton I, Fernandez JM, Sala G.
Stefancich from the Universitá di Ferrara, for their help in
the installation and setting up of the three new•UCLID•S. Laser grooved buried contact cells optimised for linear
concentration systems. Solar Energy Materials and
Solar Cells 2010; 94: 187 193.
REFERENCES
12. Vivar M, Sala G, Ąachón D, Antón I. Large area Si
1. Sala G, Arboiro JC, Luque A, Zamorano JC, Minano cell encapsulation for concentrator systems: review of
JC, Dramsch C, Bruton T, Cunningham D. The critical points and new proposal for the third genera-
•UCLID•S prototype: an efficient parabolic trough for tion of •UCLID•S. 4th IEEE World Conference on
Ä„V concentration. 25th IEEE Photovoltaics Specialists Photovoltaic Energy Conversion, Hawaii, USA, 2006.
Conference, Washington, USA, 1996; 1207 1210. 13. Vivar M, Antón I, Ąachón D, Sala G. Type approval of
2. Antón I, Sala G. The •UCLID•S concentrator, Chapter •UCLID•S III receivers using the I•C 62108 norm
13. In Concentrator Photovoltaics, Luque A, Andreev Draft 8 intended for Ä„V concentrators. 21st European
V (eds). Springer: New York, 2007. Photovoltaic Solar Energy Conference and Exhibition,
3. Luque A, Sala G, Arboiro JC, Bruton T, Cunningham Dresden, Alemania, 2006.
D, Mason N. Some results of the •UCLID•S photo- 14. Vivar M, Herrero R, Antón I, Martínez Moreno F,
voltaic concentrator prototype. Progress in Photovol- Moreton R, Sala G, Blakers AW, Smeltink J. •ffect
taics: Research and Applications 1997; 5: 195 212. of soiling in CÄ„V systems. Solar Energy 2010; 84:
4. Sala G, Arboiro JC, García R, Luque A, Bruton T, 1327 1335.
Cunningham D. Description and performance of the 15. Sala G, Ąachón D, Antón I. Modelling of the reference
•UCLID•S concentrator prototype. 14th European concentrators, C RATING Final Report, Book 2, 2003.
Photovoltaic Solar Energy Conference, Barcelona, 16. Vivar M, Antón I, Ąachón D, Sala G, Ąachón JL,
Spain, 1997; 352 355. Álvarez R, Montalvo F. Third generation of
5. Sala G, Arboiro JC, Luque A, Anton I, Gasson MÄ„, •UCLID•S systems: first results and modelling of an-
Mason NB, Heasman KC, Bruton TM, Mera •, nual production in ID•OCONT• project test sites. 4th
Camblor •, Datta Ä„, Cendagorta M, Friend MÄ„, Velera International Conference on Solar Concentrators for
Ä„, Gonzalez S, Dobon F, Ä„erez F. 480 kW peak the Generation of Electricity or Hydrogen, •l •scorial,
•UCLID•S"! concentrator power plant using para- Spain, March 2007.
bolic troughs. 2nd IEEE World Conference on Photo- 17. Vivar M, Herrero R, Martínez Moreno F, Antón I,
voltaic Energy Conversion, Vienna, Austria, 1998. Sala G. •ffect of soiling in Ä„V concentrators: mech-
6. Sala G, Antón I, Arboiro JC, Luque A, Camblor •, anisms of light dispersion and real field performance
Mera •, Gasson MÄ„, Cendagorta M, Valera Ä„, Friend of soiled flat modules and CÄ„Vs. 23rd European Pho-
MÄ„, Monedero J, González S, Dobón F. The 480 tovoltaic Solar Energy Conference and Exhibition,
kWp •UCLID•S"! TH•RMI• power plant: installation, Valencia, Spain, September 2008.
set up and first results. 16th European Photovoltaic Solar 18. Real Decreto 436/2004 of 12th of March of 2004,
Energy Conference, Glasgow, UK, 2000; 2072 2077. BO•, 2007.
Prog. Photovolt: Res. Appl. 2012; 20:356 371 © 2011 John Wiley & Sons, Ltd. 371
DOI: 10.1002/pip