Session 3A 1

SESSION 3A VISUAL PRESENTATIONS
4th International Conference on Solar Concentrators for the Generation of Electricity or Hydrogen
COMPARISON BETWEEN DIFFERENT SOLAR CONCENTRATORS AS REGARDS TO THE ELECTRIC GENERATION

Franco Rasello, Emanuele Renzetti INTEGRARE (INTEGRA Renewable Energy)
Abstract
In this article we compare different solar concentrator technologies available in the market of the power generation. There are two main technologies that use solar energy: solar thermal concentrator (STC); photovoltaic concentrator (CPV).CPV are then divided into two classes: low and high concentration (LCPV & HCPV). The question is: which is the best technology to realise a project for solar energy conversion in relation to the site, electric generation and application type. In the first solution, STC, the total cost is divided between thermal collector, tracking, power block and HTF (Heat Transfer Fluid) system. This technology is regulated by economy of scale; for an application of 1 MW its not competitive with CPV but we have better results with a bigger size. In the second solution, CPV, the total cost is divided between solar cell, concentrator, tracking and inverter. The LCPV that uses low concentration factor could be a good compromise. We can reduce the cost of the cells, in relation to the concentrator factor. Its even possible to use Silicon cells that are ready now instead of more complex, expensive and not market ready solutions. The HCPV using high concentration factor needs to use more efficient cells (like MJ) to be efficient, but these technologies are very expensive and not mass market (at the moment).Another important difference between the two solutions (LCPV vs. HCPV) is that a low concentration factor needs a low precision tracking system and doesn't reach high temperature (low costs).A HCPV needs a high precision tracking system (higher costs) and reaches high temperature on the cells.
radiation like Decimomannu in the south of Sardinia (the irradiation values are based on 5 years of collection data). The data in this paper are obtained from several publications and some are extrapolated. [1] This is due to the fact that solar concentrators are entering now the market (especially HPVC) and so there isnt a big installation history to analyse.
2.
Solar technologies description
SOLAR TERMAL CONCENTRATION (STC) Solar thermal electric systems need an intermediate thermodynamic cycle to produce electricity. The state of the art of this technology is a great power plant size >300MW, with a molten salt as Heat Transfer Fluid (HTF) and vapour as working fluid. Solar thermal power plant could be divided into three different parts: (a) Solar field (thermal collector) (b) HTF, that includes the fluid and all aspect relative to the storage system and heat exchange (c) Power Block that includes turbine, cooling tower, heat recuperator and generator. The analysis of a smaller power plant size (1 MW and 5 MW) is only to compare this power plant with PV technologies. [2] For this small size, steam Rankine cycle isn't the best solution so we have analysed an Organic Rankine cycle derived from geothermal. The efficiency of the power block is 32% for an ORC cycle. For a steam Rankine of a 1MW size the efficiency is lower, as it decreases with dimension.(The efficiency of an advance steam Rankine cycle for power plant >150 MW is 43%).[3]
1. Introduction
The objective of this work is to determine the best solar technology that can be used in relation to the generated power and which component can be improved to obtain better price. The results could be important both for designers and for producers. To realise a commercial solution the most important thing is to reduce the cost of the system and not only cell efficiency. [6] To compare the cost of these technologies we are considering the cost of energy production (not of the Wp); to do so we consider a place with direct solar
Fig. 1 Functional diagram of a STC solution.
107
This is a complex technology, it requires high operation & maintenance cost and a continuous presence of operators. [4]The main advantage related to this technology is the possibility to store heat to assure availability even during night hours. The performance of this plant is 22, 4 % for a 1MW installed and 24, 1% for 5MW. This is given by product of
High PV Concentrator (HCPV)

The second system analysed is a high concentrator module (500X). This represents a solution available on the market , with a tandem GaAs solar cell that reaches a 32% efficiency; several studies show that the III-V J solar cell could have better performance (more then 40% ).[7] Chip size of the useful cell is 1 mm^2; the material used is in reality 60% greater so the total area becomes 1, 69 mm^2. The cost of tandem cell considered is 14, 7 /cm^2 [6]; referred to the cost / surface area of concentrator it becomes 0, 0294 /cm^2. The module efficiency is given by the product of three efficiencies: module = cell * optics * electronic We reach a value of 24, 8% with GaAs, and 31 % with MJ. The annual system efficiency of the power plant is 18, 5% for the first configuration and 23, 1 % for MJ solution. Operation and maintenance costs are the ones of the structure, inverter and tracking, The system could be completely automatic and could be monitored by a remote control. No person is necessary to control this power plant. Finally we find the energy cost, [/kWh]. For the first solution we have 0,147 /kWh and 0,131 /kWh for MJ. Table 2. Cost of the system components in %.
Technology Nominal power [MW] Concentrator solar celll [/Wp] Optics and encapsulation [/Wp] Cooling [/Wp] Structure , tracking , assembling and DC wiring [/Wp] Land and Preparation [/Wp] Inverter , transformer , electrics [/Wp] Total cost [/Wp] O&M [/Wp-year] Energy production [Wh/m2-yr] Energy cost [/kWh] HCPV 500X [GaAs] 1 1,316 0,184 0,134 51,1 7,2 5,2 % HCPV 500X [MJ] 1 1,369 0,147 0,107 57,6 6,2 4,5 %
field = mirror * htf * power cycle

In table 1 are displayed the components of the total cost. Table 1. Cost of the system components in %.
Tecnology Nominal power [MW] Solar field [/Wp] HTF [/Wp] Power Block [/Wp] auxiliary [/Wp] land and preparation [/Wp] total investement [/Wp] O&M [/Wp-year] STC 1 0,791 0,061 1,893 0,120 0,396 3,261 0,100 24,3 1,9 58,1 3,7 12,1 100,0 3,1 %cost 5 0,712 0,055 1,097 0,076 0,089 2,030 0,088 24,10% 322 0,14 35,1 2,7 54,1 3,7 4,4 100,0 4,3 %cost
System efficiency 22,50% Energy production 300 [Wh/m2-yr] Energy cost [/kWh] 0,18
The cost of electricity generated is calculated for two different plants; in fig.3 its clear that the dimension of the plant influences the cost of the power block.
PV concentrator technologies
There are two main benefits from the use of concentration in PV systems: (a) the increase in efficiency and (b) the decrease in cost. We analyse two systems based on CPV technology LCPV and HCPV. The main difference between LCPV and HCPV is the concentrator factor and the types of solar cells used. Optics is a TIR-R lens that reaches high concentrator factor, with a good acceptance angle. [5].Another important aspect is the thickness of the module. The cost is in relation to the number of lens per m^2 and this is obtained from real data of optoelectronics industry. The formula used is 1600/N^0, 5 + 0,027N (optics cost + encapsulated cost in /m2). N is the number of optical concentrators per m^2. [6] The cooling system is an important aspect of the module; we consider a passive cooling (an active cooling is the best solution to increase the efficiency but more complex technology adds more cost to the total). [6] Other costs considered are due to the module assemblage, mounting of it upon the structure, tracking systems (single axis or two axes), inverters, cables etc. These values are taken from existing power plants.
0,626 0,089 0,224 2,574 0,045 268 0,147
24,3 3,5 8,7 100 1,7
0,501 0,072 0,179 2,375 0,036 335 0,131
21,1 3,0 7,5 100 1,5
Low concentrator (LCPV)

The third system considered has a low concentrator factor. We analyse three different solutions; a very low concentrator factor with 20X, 100X and 200X.
108
The 20X solution doesn't require a high precision angle of acceptance; so if C < 20X is possible to use a single axis tracker (this decreases both the cost of the structure and the tracker). With this LCPV its possible to realize solutions to be installed on the roof and in building integration (like flat plat module). The cell type used for low C is a Silicon solar cell with a rear contact. Several studies of this cell under a C factor show that a non linear efficiency improvement is possible. We optimized the dimension of the chip in relation to the C factor; for a 20X the best solution its a 5mm square cell, for 100X and 200X its 2mm. The maximum cell efficiency is 27% for 100X and its 26, 5% for 200X and 20X. [8] The cost of silicon cell is more then 150 times lower than GaAs and MJ. It's about 0, 09 /cm^2, so the impact on the total cost of the system is low. This is a very important aspect because the cost for Silicon, in a photovoltaic standard technology is 60 % of the total cost. Module efficiency is 21, 3 % for a 20X, 21, 5% for 100X and 21, 3 for 200X . Operation & maintenance cost are similar to the previous system, but using a single axis tracking system the cost decreases. Table 3. Cost of the systems components in %.
Tecnology Nominal power [MW] Concentrator solar cell [/Wp] Optics and encapsulation [/Wp] Cooling [/Wp] Structure , tracking , assembling and DC wiring [/Wp] Land and Preparation [/Wp] Inverter , transformer , electrics [/Wp] Total cost [/Wp] LCPV 20X [5mm] 1 0,229 0,398 0,153 13,9 24,3 9,3 % LCPV 100X [2mm] 1 0,045 0,392 0,152 3,1 27,1 10,5 % LCPV 200X [2mm] 1 0,023 0,446 0,152 1,56 29,76 10,12 %
Fig. 2 represents two solar thermal plants (STC) with different size, and same technology .
STC solar thermal concentrator
0,20 0,18 0,16 0,14 O&M LAND OTHER COST POWER BLOCK HTF SOLAR FIELD 1MW 5MW
/ kWh
0,12 0,10 0,08 0,06 0,04 0,02 0,00
Fig.2 Cost of the system components in % The main cost is given by the solar field and the power block (which includes turbine, condenser, and heat exchanger). If we compare a 5 MW and 1 MW plant we observe that the power block cost decreases and the solar field increases (Fig.2). The main reason is the improvement of power block efficiency. In Fig.3 we ca see how plant size affects the energy cost of this type of technology.
STC curve cost

0,2 0,18 0,16 0,14
/kWh
0,12 0,1 0,08 0,06 0,04
0,559 0,102 0,200 1,641
34,1 6,2 12,2 100,0 3,1
0,556 0,101 0,200 1,445 0,051 237 0,113
38,4 7,0 13,8 100,0 3,5
0,573 0,104 0,200 1,498 0,052 230
38,25 6,95 13,35 100,00 3,48
0,02 0 0 50 100 150 200 250 300 350 P ow e r [M W ]
O&M [/Wp-year] 0,051 Energy production 235 [Wh/m2-yr] Energy cost [/kWh] 0,115
Fig 3 Cost of the system vs. installed power

0,114
3. Cost comparison
To compare different solar technologies, we use, as main parameter, the cost of energy generated (it includes all capital costs, energy production, O&M cost, and financial aspect). Every item of the total cost is referred to the energy cost in order to analyse which component of the system is the most important and could be improved.
For HCPV the main cost is the solar cell concentrator and tracking. In Fig. 4 we have the energy cost for two modules using tandem solar cell and MJ, with 500X and 1000X Concentration level. The C factor has a big influence on the cell cost in relation to the entire module; this is 33% of the cost of a 1000X, and 51% for a 500X solution. The MJ solution is a new technology and so we can expect a good optimisation of it in the near future.
109
HCPV
0,16 0,14 0,12 0,10 INVERTER, TRANSFORMER ,ELECTRICS LAND and PREPARATION STRUCTURE, TRACKING,ASSEMBLING and DC WIRING COOLING OPTICS and ENCAPSULATION CONCENTRATOR SOLAR CELL 1000X 500X 1000X 500X OPERATION & MANTEINANCE
0,08 0,06 0,04 0,02 0,00
Concentrar factor [GaAs]
[MJ]
Fig. 4 Cost of the systems components in %. In the LCPV study we consider three different concentrator factors and 4 different chip sizes (which as we see in Fig. 5 are important to optimize the number of optics and their costs).
This technology is under improvement and has the possibility to work even in dark condition using storage heat tanks. For a middle size plant ( 100KW to 10-50 Mw) with direct solar radiation the best solution could be HCPV especially if MJ will decrease in price and increase in efficiency (as its happening right now). A lot of improvements are under development and we are expecting in 2007 the start of the market. The LCPV technology could be interesting in the small power plant and in the near term. It seems interesting to use, for their production, the same LED technology to reduce the cost. These systems may have similar dimension as a 1 sun PV and so they could be used even on roof solution or BIPV using a single axis tracking. The LCPV solutions that are using current Silicon production should use in the future new MJ as their price will decrease.
/kWh
References
[1] http://re.jrc.ec.europa.eu/pvgis/pv
LCPV
0,25
0,20
OPERATION & MANTEINANCE INVERTER, TRANSFORMER ELECTRICS LAND and PREPARATION STRUCTURE, TRACKING
0,1 5
0,1 0
COOLING
0,05
[2] E. Prabhu,California Solar Trough Organic Rankine Electricity System (STORES) Stage 1: Power Plant Optimization and Economics November 2000 May 2005 Reflective Energies Mission Viejo, California NREL/SR550-39433 March 2006. [3]M.Falchetta Il programma Enea sull'energia solare a concentrazione ad alta temperaturaENEA Grande Progetto Solare Termodinamico Unit Ricerca e Sviluppo Centro Ricerche Casaccia Via Anguillarese, 301- S.Maria di Galeria 00060 Roma [4]D.Kearney , H.Price (2006) Advances in solar power plant technology an annual review of research and development pp 204-223 [5] Jose L.Alvarez , Vincente Diaz , Jusus Alonso Optics design key points for high gain photovoltaic solar energy concentrators ISOFOTON s.a Severo Ochoa ,50 ,Parque Tecnologico de Andalucia Malaga 29590 Spain [6]A. Mart, A. Luque Next generation photovoltaics, high efficiency through full spectrum utilization chapter 6 pp 108-133 [7]R. McConnell, M. Symko-Davies Multijunction Photovoltaic Technologies for High-Performance Concentrators Presented at the 2006 IEEE 4 World Conferences on Photovoltaic Energy Conversion (WCPEC-4) Waikoloa, Hawaii May 712, 2006 [8] A.Mohr aus Stegen Silicon Concetrator Cells in Twostage photovoltaic system with a concentrator factor of 300X Dissertation zur Erlangung des Doktorgrades der Fakultt fr Angewandte Wissenschaften der AlbertLudwigs-Universitt Freiburg im Breisgau [9] ISOFOTON price
/kWh
0,00
OPTICS and ENCAPSULATION CONCENTRATOR SOLAR CELL
cell size [mm] concentrator factor [20 100 200]
Fig. 5 Cost of the systems components in %. The most influential components are structure, tracking and optics and encapsulation which can be reduced using LED technologies and by moving forward on the learning curve. Concentrator solar cell cost is 12% of the total for a very low C factor (20X) , 2% of a 100X and only 1% of a 200x. This means that high concentration level is not an optimal solution for silicon cell. However LCPV can be used with MJ cells as soon as their cost will decrease; recent studies show that some MJ cells have high efficiency at C equal to 5-10X [7]. The thickness of low LCPV could be very small (similar to the 1 sun PV).This aspect, with the use of a single axis tracking system, give to the LCPV technology the possibility to be adopted in BIPV or roof solutions with a cost three times smaller then a 1 sun PV.
4. Conclusion
From this study we can make the following consideration. To realize a big power plant (>10-50MW) the most efficient technology for best ROI is STC (Solar Thermal Concentrator) using different typologies as Parabolic Mirrors, Solar Tower or Dish.
110
EXPERIMENTAL TEST AND MODELLING OF CONCENTRATOR SOLAR CELLS UNDER MEDIUM AND HIGH FLUXES
A. Vossier1, S. Quoizola2, S. Grillo2, G. Flamant1 and A. Dollet2* Laboratoire PROMES-CNRS 1 B.P. 5 Odeillo - 66125 Font Romeu Cedex -FRANCE 2 Tecnosud - Rambla de la thermodynamique, 66100 Perpignan France *Corresponding author: phone +33 4 68682212, e-mail: dollet@univ-perp.fr ABSTRACT The goal of this work was to study the potential of some concentrator solar cells for operating under high flux. Single junction GaAs cells and InGaP/InGaAs tandem cells (ISE Fraunhofer) were tested in outdoor conditions. 2 dish mirror-based systems were developed for performing measurements: a medium concentration system working below 800 suns and a high concentration system working up to 10,000 suns. Simple analytical models were derived from I-V measurements under low flux in order to evaluate the conversion efficiencies () at various concentration ratios (X) and temperatures (Tcell). By modelling heat transfer in the CPV systems, 2D temperature profiles in the cells were calculated for various concentration and cooling conditions. Theoretical values of (X, Tcell) were compared to values measured under medium flux conditions. While free convection on a Cu plate is sufficient for cooling cells under medium concentration, active cooling becomes necessary at very high concentration. INTRODUCTION Concentration of sunlight is known to be a promising way of reducing the cost of photovoltaic conversion. To date, most concentrator solar cells have been studied under low and medium concentration ratios (below 500 or 1000 suns), but there is little literature available today on solar cells operating under very high fluxes (between 1,000 and 10,000 suns) [1]. At such high concentration levels, several technological requirements must be achieved: high efficiency of the cooling system, good quality optics and very low series resistance in order to minimise the electrical losses in the cell [1]. The ultimate goal of this work will be to test selected concentrator solar cells under very high solar fluxes (up to 10,000 suns) in real outdoor conditions. In this paper, as a starting point, various concentrator solar cells will be tested under medium flux, then simple simulations of heat transfer and energy conversion will be performed. From both series of results, appropriate conditions and solar cells will be selected for forthcoming ultrahigh flux experiments.
EXPERIMENTAL PROCEDURE Experiments were conducted at Odeillo, a place in the south of France where particularly good direct sunlight conditions can usually be found. Solar facilities from either the DGA-CEP or PROMES laboratory have been used. The parabolic dish used for medium concentration experiments (figure 1) has a diameter of 1 m and a focal length of 2.8 m. The focal spot diameter is about 30 mm. The maximum concentration ratio is approximately 800 at the center of the spot. The concentration level is modulated by means of an iris placed in front of the dish.
Fig. 1. Schematic view of the medium concentration experiment (heliostat+ dish double reflection system) The parabolic dish used for very high flux experiments has a diameter of 1.5 m and a focal length of 0.65 m. The focal spot diameter is 16 mm. The maximum concentration ratio is approximately 10,000 at the center of the spot.
Concentrated beam Cooled mask Glass rod Solar cell Heat sink
Fig. 2. Schematic view of the cooled mask and optical guide designed for high concentration experiments
111
In both systems, a water-cooled mask with a 2mm hole in its centre is placed above the cell in order to fit the size of the incoming beam to the 2 mm cell, and avoid any overheating of the surrounding parts (e.g. electrical contacts). In the particular case of the high concentration system, which has a short focal length, the concentrated beam is transmitted through a silica rod that is fixed to the water-cooled mask (fig. 2). In both systems, the solar cell is mounted on a copper plate, itself mounted on a watercooled support. IV curves were recorded with a Keithley 2600 sourcemeter. EXPERIMENTAL RESULTS Our very first experimental tests were performed under medium concentration (1<X<600) on GaAs and GaInP/GaInAs cells delivered by the ISE Fraunhofer Institut. Silicon concentrator cells from ISE have been also tested but III-V solar cells exhibited superior efficiencies, especially for concentration ratios > 100. GaAs and tandem cells have the following characteristics under 1 sun (AM 1.5): Voc=0.961V, Isc=0.81 mA, FF=0.806, =17.9% for the GaAs cell and Voc=2.002V, Isc=0.469 mA, FF=0.864, =22.5% for the GaInP/GaInAs cell. Figure 3 shows a series of experimental I-V characteristics recorded on the tandem cell. It is worth noting that the slope of current vs voltage curves exhibit slight changes around Voc (for both the GaAs and the tandem cells). This rather unusual behaviour has already been observed under high concentration conditions by Siefer et al. [2] and ascribed to a Schottky contact. As seen in fig. 4, this behaviour is no longer observed for X<100.
0 -0,02 -0,04 -0,06
0 0,5 1 1,5 2 2,5
We have also observed that the magnitude of this effect decreases with decreasing cooling efficiency (that is, with increasing cell temperature). Plotted in Figure 5 and 6 are the conversion efficiency vs concentration ratio (X) for the GaAs cell and tandem cell respectively. The highest efficiencies are observed for X100 (GaAs) or 150 (tandem); however, they should theoretically be obtained for greater values of X (between 200 and 500). Possible explanations are that our measured efficiency have not been corrected for cell temperature increase under concentration, or that uncertainties exist on Voc measurements due to the Schottky contact.
Fig. 5: GaAs cell conversion efficiency vs concentration ratio (efficiency was calculated by accounting for optical losses in the system).
27% 26%
efficiency (%)
25% 24%
X=108 X=167 X=242 X=310 X=383 X=445 X=518 X=588
cell efficiency (from est. optical losses)

23% 22% 21% 0 100 200 300 400 500 600
current (A)
-0,08 -0,1 -0,12 -0,14 -0,16 -0,18 -0,2
concentration ratio (X)
Fig. 6: GaInP/GaInAs cell conversion efficiency vs concentration ratio (efficiency calculated by accounting for optical losses in the system)
voltage (V)
Fig. 3. I-V plots recorded for medium concentration ratios (100<X600) on 2 mm GaInP/GaInAs cell from ISE (direct -2 sunlight:890Wm )
0 0 X=12.5 -0,005 X=1 0,5 1 1,5 2 2,5
current (A)
-0,01 X=40 -0,015
FF factor values remain good for both cells (0.7-0.8 and 0.8-0.88 for the GaAs and tandem cell respectively) in the concentration rage investigated, but slightly decrease up to approximately 100 suns. The above results suggest that both GaAs and tandem cells tested in this work may operate under slightly higher concentration, but FF and efficiencies might decrease more significantly for very high values of X. HEAT TRANSFER MODELLING Avoiding material overheating while exposing cells to very high solar fluxes is a very important requirement for high concentration photovoltaics. The main consequence of an excessive increase of cell temperature is a reduction of the open-circuit voltage, and consequently a drop in the maximal electrical power delivered by the system. Passive systems based on natural convection (air) are routinely used today for cooling solar cells exposed to
-0,02 X=73 -0,025
voltage (V)
Fig. 4. I-V plots recorded for low concentration ratios (X<100). Cell type and conditions as in fig. 3
112
several hundreds of suns. However, operating cells with concentration ratios between 1,000 and 10,000 requires very efficient active cooling systems for dissipating the non-converted fraction of absorbed radiation as well as the heat due to increased ohmic losses at high current densities. In order to estimate the temperature increase in the cell under various concentration ratios, prior to conducting experiments, 2D simulations of heat transfer inside the cell and Cu heat sink were performed by using the COMSOL (FemLab) software. In the case of passive cooling, heat is transferred by conduction in the solid parts and extracted by radiative and convective exchanges between the solids and ambient air, while in the case of active cooling, an additionnal "forced convection" contribution is added (water-circulation). The conduction part of the heat transfer writes, in steadystate conditions:
The reference diameter of the Cu plate (assumed to be circular in this 2D simulation) is 20 mm. However, it was not possible to keep the cell temperature at a reasonably low level at very high concentration by using such a small heat sink cooled by free convection only. Hence, the Cu plate surface was allowed to vary linearly with the concentration ratio (the area of the Cu plate was equal to the product of the active area of the cell by the concentration ratio X). A solar cell temperature of 80C has been defined as the maximum temperature that must not be exceeded in order to prevent a possible degradation of the cell connections and encapsulation [3]. Table 1 shows the steady state cell temperatures obtained by simulation for various concentration ratios. Therefore, passive cooling may not be sufficiently efficient to maintain a low cell temperature at very high concentration ratios. 20 63 141 200 Cu plate diameter (mm) Concentration ratio 100 1000 5000 10 000 (X) Cell temperature 62 87 110 136 (C) Table 1 : Temperature of a 3.1 mm GaAs solar cell under concentrated sunlight (passive cooling). Case 2 : Active cooling In the case of active cooling, the heat transfer coefficient at the back side of the heat sink is significantly higher than for natural convection in air. The average forced convection heat transfer coefficient is given by:
2T x 2
= Qcond
(1)
where k is the thermal conductivity of the solid (cell or heat sink). Radiative exchanges between the cell and ambient air are described by the following relation:
Qray= (T4-T04)
(2)
where is the emissivity of the material and the StefanBoltzmann constant. Values of and k can be easily found in handbooks. Passive cooling and active cooling require two different treatments for convection: Case 1 : Passive cooling The heat flux exchanged by natural convection can be written as [3] : (3) Qconv = h(Tcell -Tair) where h is the natural convection coefficient. The natural heat transfer coefficient of the upper (front) surface (cp-1) of the heat sink is given by :
hp =
where
Nu p k f Dp
and
(9)
Nu p =1 Re2 Pr1/ 3
Re= u f Dp vf
(10)
where 1 and 2 are tabulated values. In the above expressions, uf is the cooling fluid mean velocity and f is the kinematic viscosity of the fluid. 100 1000 5000 10 000 Concentration ratio (X) Cell temperature 27.5 31.5 49.5 72 (C) Table 2 : Temperature of a 3.1 mm GaAs solar cell under concentrated sunlight (active cooling). The Cu plate diameter is 20 mm. As seen in table 2, the cell temperature never exceeds 80C even at concentration ratios of 10,000. As a consequence, efficient active cooling systems must be used to operate solar cells under very high fluxes. EFFECTS OF SERIES RESISTANCES
hcp 1 =
where
Nu cp 1k f Dp
(4)
1/ Nucp 1 =0.59Racp4 1
Racp 1<109 109 < Racp 1
(5) (6)
Nucp 1 =0.10Ra1/ 31 cp
and for the rear surface (cp-2) of the heat sink :
hcp 2 =
where
Nu cp 2k f Dp
(7)
1/ Nu cp 2 =0.27Racp4 2 105 < Racp 2 <1011
(8)
In the above expressions, kf is the thermal conductivity of the fluid (air), Dp is the plate diameter. Ra and Nu are the well-known dimensionless Rayleigh and Nusselt numbers.
The most limiting factor for photovoltaic conversion under very high solar flux might be due to series resistances, and particularly the power dissipation by joule effect, which is proportional to the second power of the current flowing into the electrical circuit. The series
113
resistances (Rs) of concentrator solar cells are usually small (typically in the range 1-10 m), but additional resistances due to undesirable bad electrical contacts in the circuit, as well as the resistance of the electrical wires must be taken into account. For instance, considering the size of our dish and the fact that the I-V measurement system is located approximately 1.5 m away from the focus plane, a 3 m long electric wire with a non-negligible resistance is needed to connect the cell to the Keithley sourcemeter. The resistance Rw associated with this electric wire can be estimated by the following relation : Rw =
100% 2.34% 23.3% 5000 100% 4.6% 45.9% 10 000 Table 3 : Evolution of the ratio of power loss for a 3.14 mm GaAs solar cell under high solar fluxes. Ratio of power loss (%) Concentration R= 10 m R= 100 m R= 1 ratio 100% 3.6% 35.8% 1000 100% 7.03% 70.3% 2000 100% 100% 17.2% 5000 100% 100% 33.8% 10 000 Table 4 : Evolution of the ratio of power loss for a 16 mm GaAs solar cell under high solar fluxes. Ratio of power loss (%) Concentration R= 10 m R= 100 m R= 1 ratio 100% 100% 22.3% 1000 100% 100% 43.95% 2000 100% 100% 100% 5000 100% 100% 100% 10 000 Table 5 : Evolution of the ratio of power loss for a 100 mm GaAs solar cell under high solar fluxes. Finally, we have performed a 2D simulation with ComSol in order to simulate the temperature increase due to joule effect in the small electrical gold connections extracting the current from the cell core. The temperature of the gold wires may exceed 900C at 10,000 suns in our cells, which have typically only 4 gold wires. It is concluded that our solar cells could perhaps operate at slightly higher concentration, but smaller cells with a rather large number of electrical gold wires must be used in order to limit the temperature increase in the cell or in electrical contacts under very high concentrations. ACKNOWLEDGEMENTS We would like to thank J. Gordon, from Ben Gurion University (Isral) for his encouragements and stimulating discussions. We also thank R. Garcia and J-J. Huc (PROMES) for their assistance in the design of the experimental set-up, J-J. Serra, JM Sayous and E. Scheer (DGA) for invaluable technical support. This work was partly conducted in the framework of an agreement between DGA and CNRS (MCTS research team). REFERENCES
[1] C. Algora. The importances of the very high concentration in third generation solar cells. Next generation photovoltaics: high efficiency through full spectrum utilization 2004. pp. 108-136. [2] J. Sun, T. Israeli, T. Agami Reddy, K. Scoles, J. Gordon and D. Feuermann. Modelling and experimental evaluation of passive heat sinks for miniature high-flux photovoltaic concentrator. Journal of solar energy engineering 2005. 127, . pp. 138-145. [3] G. Siefer, P. Abbott, T. Schlegl and A.W. Bett. Determination of the Temperature Coefficients of Various III-V Solar Cells. 20th European Photovoltaic Solar Energy Conference 2005.
l
S
(11)
where , l and S are respectively the resistivity, the length and the cross-sectional area of the wire. Of course, the value of the resistance of this electric wire should be minimized by choosing a very low resistivity metal, such as silver, and by lowering the l/S ratio. The total resistance of the circuit is equal to the sum of the resistance associated with the electric wires Rw and the series resistance of the cell Rs. (12) R = Rw + Rs Three values of the circuit resistance have been considered here, in order to evaluate the power losses in the circuit when the total resistance is low (R=10 m), moderate (R=100 m) and very high (R=1).. The ratio of power loss is defined as the power dissipated by joule effect divided by the power delivered by the cell at the maximum power point (Pcell) : rPL=
Ploss Pcell
(13)
In our case, the power delivered at the maximal power point has been simply calculated from the measured value at 1 sun, corrected for the theoretical voltage increase resulting from the concentration, but losses due to series resistance have been neglected in this calculation, for a simplification purpose only. This means that the values reported in this paper for the ratio of power loss are lower limits. This ratio has been calculated for three different sizes of solar cells : 3.14, 16 and 100 mm, and for four different concentration ratios (1,000; 2,000; 5,000 and 10,000 suns). Table 3 to 5 show that power dissipation by joule effect may drastically reduce the power delivered by the larger cells submitted to very high concentration ratios. The most critical situation corresponds to large cells and/or large resistances. In this case, all the power delivered by the cell is dissipated by joule effect. Very mall size solar cells may efficiently work, even at quite high solar fluxes, if the circuit resistance is kept very small (less than 10 m). Ratio of power loss (%) R= 1 R= 10 m R= 100 m 0.5% 0.96% 4.9% 9.6% 48.6% 95.5%
Concentration ratio 1000 2000
114
HIGH-FLUX CHARACTERIZATION OF ULTRA-SMALL TRIPLE-JUNCTION CONCENTRATOR SOLAR CELLS

Omer Korech, Baruch Hirsch, Eugene A. Katz and Jeffrey M. Gordon
Department of Solar Energy and Environmental Physics, Jacob Blaustein Institutes for Desert Research, Ben-Gurion University of the Negev, Sede Boqer Campus 84990, Israel
ABSTRACT
Ultra-small multi-junction solar cells comprise the latest generation of commercial concentrator photovoltaics, with claims of cell efficiency ~40% achieved at several hundred suns. Cell miniaturization ostensibly allows greater peak efficiency at higher flux, facilitates passive heat rejection and permits the practical use of allglass optics. However, few measurements have been reported, in particular as a function of concentration and flux distribution. We present extensive measurements on commercial ultra-small triple-junction solar cells with a solar fiber-optic mini-concentrator. Flux maps on the 1.0 2 mm active region within the busbars were varied from strongly inhomogeneous to uniform, at delivered flux levels up to ~5000 suns. These results allow assessments of (a) optical and internal resistive losses, as well as (b) cell performance at high flux, including (c) sensitivity to the type of flux inhomogeneities encountered in high-flux optical devices.
performance for broad ranges of anticipated operating conditions.
EXPERIMENTAL
All experiments were performed with a dual-axis tracking mini-dish solar concentrator [8] wherein transmissive (quartz-core) optical fibers 1.0 and 0.6 mm in diameter channeled concentrated sunlight to an indoor test bench (Fig 1). The localized irradiation probe (LIP) [10-11] delivered light confined to the circle delimited by the fiber tip. This procedure can simulate the intensity and type of non-uniform distributions produced by many practical high-flux optics [3, 5-6]. Uniform irradiation of the cell's active area was produced with a square glass kaleidoscope coupled between the distal fiber tip and the cell (Fig 1b). Solar irradiation on the cell Pin was moderated with an iris (Fig 1a) and measured pyrometrically. LIP levels were varied continuously. The highest localized concentration with the 0.6 and 1.0 mm fibers was 3700 and 5100 suns, respectively.
INTRODUCTION
Multi-junction solar cells have already demonstrated conversion efficiencies above 40% at 2 several hundred suns [1] (one sun 1 mW/mm ). High concentration with inexpensive high-flux optics allows reducing the system share of costly photovoltaic (PV) cells. Practical considerations then motivate cell miniaturization, which can [2-4]: increase cell efficiency, raise the flux value at which efficiency peaks, facilitate passive heat rejection and permit all-glass compact miniaturized optics [5-6]. The latter was recently suggested for realizing net flux values in excess of 103 suns at high collection efficiency with completely passive 2 heat rejection for a cell area of ~1 mm . Such ultra-small triple-junction solar cells lie at the core of the latest generations of commercial concentrating PV systems. However, there are few published data on cell performance, in particular as a function of concentration and flux distribution. Here we present measurements on commercial ultra-small triple-junction GaInP2/GaAs/Ge solar cells [7], generated with our ultrahigh flux real-sun fiber-optic minidish concentrator [8-11], and elucidate the principal cell parameters essential for appraising concentrating PV
Fig 1. (a) Solar mini-concentrator (20 cm diameter) with fiber-optic transmission indoors [8]. Input power is moderated by an iris that preserves the angular distribution of delivered sunlight. (b) Uniform cell irradiation by a kaleidoscope. (c) Fiber-cell contact in the LIP mode.
115
Measurements were limited to clear-sky periods, two hours about solar noon, over the course of several months in Sede Boqer, Israel. The light spectrum on the cell was nearly invariant and close to the air mass 1.5 direct beam solar spectrum [8]. Cell dimensions are indicated in Fig 2. Currentvoltage (I-V) curves were measured with 3 patterns: (a) 2 uniform illumination of the full 1.0 mm active area within the busbars, (b) LIP with the 1.0 mm fiber, and (c) LIP with the 0.6 mm fiber. Cells were mounted on a 1.6 mm thick gold-coated Kovar ceramic substrate which was thermally bonded to a passive copper heat sink. The maximum temperature at the heat sink-substrate interface was only 3 K above indoor ambient (actual cell junction temperatures cannot be measured directly and exceed that of the interface).
flux level of ~1,000 suns (Fig. 3b) - considerably higher than the ~350 suns at which the efficiency of an earlier 2 100 mm of the same nominal cell architecture peaked [12].
0.90
0.85
ABSTRACTS
Fill Factor
Position the word ABSTRACT (all upper case) 0.5 (120.80 below the last line of the organization, centered in mm) the left-hand column. A blank line should be inserted between the title and text of the abstract. The abstract text is limited to 2.5 (63.5 mm). Begin the main text of the Kaleidoscope 0.75 manuscript 0.39 (10 mm) below the end of the abstract. LIP: 1.0 mm fiber
a
0.70
LIP: 0.6 mm fiber

HEADINGS
Efficiency [%]
This sheet has been generated in accordance with 10 100 1000 10000 the style to be followed for the headings. Major headings are to be in capitals without underlining, centered over one Pin [mW] column and bold. Subheadings are to be lower case with 36 initial capitals and bold. Subheadings should start at the K a le left-hand margin on id o s c o p e lines. Sub subheadings are separate 35 L treated the sameIP : 1subheadings. A blank line should be as m m fib e r L IP : 0 .6 m m fib e r placed3 4 before and after each heading or subheading. both
33 32
EQUATIONS
Fig. 2. Photograph of the square GaInP2/GaAs/Ge triplejunction cell with 4 busbars of 0.2 mm width. Net active 2 2 area = 1.00 mm . Gross area = 2.56 mm . Short-circuit current Isc was found to be proportional to Pin with Isc/Pin = 0.1380.007 A/W independent of both Pin and flux distribution. In the analyses that follow, Fill Factor FF and cell efficiency are calculated as
Equations are to be numbered consecutively throughout the paper. The equation number, in 31 parentheses, should be placed flush with the right-hand 0 margin3 of the column. When possible, use an equation editor.
29 28
b G I 2 = I 1 + I sc1 I 2 1 + [T2 T1 ] 10 1 1000 G I0 0 1

P in [ m W ]
10000
(1)
3200
= Pm/Pin = FF Isc Voc/Pin
(2)
VOC [mV]
FF = Pm/(Isc Voc)
(1)
3000
2800
where Voc denotes open-circuit voltage, and Pm is the maximum electrical power output.
2600 10
c
100
Kaleidoscope LIP: 1.0 mm fiber LIP: 0.6 mm fiber Flash simulator

1000 10000
RESULTS AND DISCUSSION

Fig. 3 summarizes all data for Voc, FF and as functions of Pin and flux distribution. Several observations are germane in evaluating such ultra-small cells for highflux applications. (A) Under uniform illumination of the active 1.0 2 mm area within the busbars, efficiency is maximum at a
Pin [mW]
Fig. 3. Dependence of (a) FF, (b) and (c) Voc on Pin for various flux distributions. The experimental uncertainties are 0.5, 2 and 5 % (relative) for Voc, FF and , respectively. In Fig 3c, the data for uniform illumination were obtained with a flash solar simulator.
116
(B) The degree of flux non-uniformity produced by LIP with the 1.0 mm fiber did not result in significant differences in I-V curves or the principal cell parameters (relative to uniform irradiation), independent of Pin, to within our experimental uncertainty of 0.5% in both I and V. (C) The cell parameter most sensitive to series resistance, and hence to irradiation distribution, FF, remains roughly independent of both Pin and flux distribution up to an input power of almost 1.0 W (Fig 3a). The impact of highly localized flux distribution (with the 0.6 mm fiber) at higher concentration is considerable as series resistance effects are expressed more prominently [11]. These trends are also reflected in the efficiency (Fig 3b). (D) By measuring I-V curves across the active 1.0 mm2 area inside the busbars with the 0.6 mm fiber, the spatial dependence of cell performance was mapped, and, for each Pin value, found to be independent of position. This reveals the homogeneous spatial distribution of both optical and series resistance losses. (The perfectly horizontal behavior of the low-voltage regime in the I-V curves we measured attested to negligible shunt losses, as in earlier investigations of cells of comparable architecture [10-11].) (E) When series resistance and heating losses are negligible, plots of Voc against ln(Pin) are linear with a slope that reveals the diode quality factor n (for an ideal triple-junction cell, n = 3)
CONCLUSIONS
This report constitutes one of the first characterization studies of the new generation of ultrasmall and nominally ultra-efficient multi-junction concentrator solar cells, with emphasis upon the sensitivity of their performance to concentration and flux distribution. Information of this type is essential in the design and optimization of new high-flux photovoltaic systems. It also highlights the value of fiber-optic mini-concentrator localized irradiation probes in mapping cell properties 2 even within a 1 mm area. Ultra-small cells are motivated in part by the prospect of higher efficiency that is realized at far higher concentration values than with previous technologies. The adequacy of the metallization and busbar design remained to be established. GaInP2/GaAs/Ge cells with an active area of 1.0 2 mm within the busbars, and busbars of essentially the same area, were probed by continuous concentrated natural sunlight (in contrast to flash simulators) at flux levels up to ~5100 suns with flux distributions (within the busbars) that varied from uniform to markedly inhomogeneous with all the light being restricted to 28% of the active area. (Passive cooling sufficed in all instances.) Their efficiency peaked at ~1000 suns for uniform 2 irradiation of the 1.0 mm region within the busbars. In contrast, earlier generations of concentrator cells had been tailored for efficiency to peak at ~200-300 suns [3,4,7,12]. The fact that plots of Voc against ln(Pin) are sublinear above ~1,000 suns indicates that an improved front contact configuration could both enhance efficiency and increase the concentration at which efficiency peaks, e.g., reconfiguring the busbar toward diminishing dark current losses. The measured diode quality factor n 6 is indicative of an excessive recombination at the cell perimeter, that could be reduced by passivation of the cell edges. The irradiation protocols and inhomogeneous flux maps used for cell interrogation can be similar to those generated in some of the high-flux optics for concentrator photovoltaics. As such, they shed light on the performance penalties (or the lack thereof) that can be anticipated in new generations of photovoltaic concentrator systems.
Voc (nkT/q) ln(Isc/Io) (nkT/q) ln(Pin) + const , (3)

where k is Boltzmann's constant, q is the magnitude of electron charge and Io denotes the reverse saturation current - consistent with our lower-flux measurements, including their insensitivity to flux distribution (Fig 3c). However, the regressed value of n is ~6, indicative of strong recombination in the depletion regions of the cell junctions. Because previous large (100 and 30 mm2) versions of the same cell architecture exhibited n 3 [1112], it would appear that the anomalously high n in these ultra-small cells stems from edge recombination, and mandates proper edge passivation in future fabrication [13-14]. (F) When flux level and/or inhomogeneity engender non-negligible series resistance losses, Voc should deviate from Eq (3) and should asymptote at high flux [11] (if cell temperature is maintained constant). When, in addition, cell temperature increases with flux, Voc may decrease as Pin is raised. Because our passive heat sinks were intended to simulate actual concentrator operation and therefore allowed junction temperatures that are non-negligibly above ambient, we filtered the temperature effect by performing measurements in a flash (s) solar simulator where the cell was uniformly illuminated and actively cooled to within 1 K of ambient (uppermost curve in Fig 3c). With heating mitigated, the effect of series resistance in lowering Voc below that of Eq. (3) becomes evident at concentration values above ~1000 suns. The signature of flux non-uniformity here likely derives from the large busbar area and the substantial dark current it contributes, which enhances the distributed character of series resistance losses [15-17].
ACNOWLEDGMENTS
We thank Vladimir Melnichak for technical assistance, Gary Conley and Steve Horne of the SolFocus Inc., Palo Alto, CA for providing the solar cells and Andreas Bett and Gerald Siefer of the Fraunhofer Institute for Solar Energy Systems for cell testing with a flash solar simulator. EAK thanks the Israel Ministry of Absorption and the Deichmann Foundation for financial support.
REFERNCES
1. US Department of Energy, Press release of December 5, 2006. New World Record Achieved in Solar Cell Technology.
117
2.
3.
4.
5. 6. 7.
8.
9.
10.
11.
12.
13.
14.
15.
C. Algora. "Very high concentration challenges of III-V MJCs," in Concentrator Photovoltaics, Eds. A. Luque and V.M. Andreev, in press (Springer, Heidelberg) 2007. A. Bett and H. Lerchemueller. "The 'Concentrix' system," in Concentrator Photovoltaics, Eds. A. Luque and V.M. Andreev, in press (Springer, Heidelberg) 2007. K. Nishioka, T. Takamoto, T. Agui, M. Kaneiwa, Y. Uraoka and T. Fuyuki. "Evaluation of InGaP/InGaAs/Ge triple-junction solar cell and optimization of solar cell's structure focusing on series resistance for high-efficiency concentrator photovoltaic systems," Sol. Energy Mat. Sol. Cells, 90 (2006) pp. 1308-1321. R. Winston and J.M. Gordon. "Planar concentrators near the tendue limit," Opt. Lett. 30 (2005) 2617-2619. www.solfocus.com/technology_gen2.html. Technical prospectus and private communications (G. Glenn), Spectrolab Inc., 12500 Gladstone Ave., Sylmar, CA, www.spectrolab.com. J.M. Gordon, E.A. Katz, D. Feuermann and M. Huleihil. "Toward ultrahigh-flux photovoltaic concentration," Appl. Phys. Lett. 84, (2004) 36423644. J.M. Gordon, E.A. Katz, W. Tassew and D. Feuermann. "Photovoltaic hysteresis and its ramifications for concentrator solar cell design and diagnostics," Appl. Phys. Lett. 86 (2005) 073508. E.A. Katz, J.M. Gordon and D. Feuermann. "Effects of ultra-high flux and intensity distribution in multi-junction solar cells," Prog. Photovoltaics 14 (2006) 297-303. E. A. Katz, J. M. Gordon, W. Tassew and D. Feuermann. "Photovoltaic characterization of concentrator solar cells by localized irradiation," J. Appl. Phys. 100 (2006) 044514. R.R. King, R.A. Sherif, G.S. Kinsey, S. Kurtz, C.M. Fetzer, K.M. Edmonds, D.C. Law, H.L. Cotal, D.D. Krut, J.M. Ermer and N.H. Karam. "Bandgap Engineering in High-Efficiency Multijunction Concentrator Cells," In: Int. Solar Conc. Conf. for the Generation of Electricity or Hydrogen, Scottsdale, AZ, May 2005, Proc. NREL/CD-520-38172 (2005). M. S. Carpenter, M. R. Melloch, M. S. Lundstrom, and S. P. Tobin. "Effects of Na2S and (NH4)2S edge passivation treatments on the dark currentvoltage characteristics of GaAs pn diodes," Appl. Phys. Lett. 52 (1988) 2157-2159. S.R. Kurtz, J.M. Olson, D.J. Friedman, J.F. Geisz, and A.E. Kibbler. "Passivation of interfaces in high-efficiency photovoltaic devices," NREL/CP-520-26494 (1999). G.M. Smirnov and J.E. Mahan. "Distributed Series Resistance in Photovoltaic Devices: Intensity and Loading Effects," Solid-State Electronics 23 (1980) 1055-1058.
16. L.D. Nielsen. "Distributed series resistance effects in solar cells," IEEE Trans. Electron Devices ED-29 (1982) 821-827. 17. V.M. Andreev, V.A. Grilikhes and V.D. Rumyantsev. Photovoltaic Conversion of Concentrated Sunlight (Wiley, Chichester) 1997.
118
FLUXMETER FOR PARABOLIC TROUGH SOLAR CONCENTRATORS

A. Parretta*1,2, M. Stefancich2, A. Antonini2, G. Flaminio3, M. Pellegrino3, L. Gentilin4, A. Maccari4, M. Montecchi4
1
ENEA Centro Ricerche E. Clementel, Via Martiri di Monte Sole 4, 40129 Bologna (BO), Italy.
2
Physics Department & CNR, University of Ferrara, Via Saragat 1, 44100 Ferrara (FE), Italy.
3
ENEA Centro Ricerche Portici, Localit Granatello, 80055 Portici (NA), Italy.
ENEA Centro Ricerche Casaccia, Via Anguillarese 301, 00123 S. Maria di Galeria (RM), Italy.
*Phone: +39 (0)51 6098617; Fax: +39 (0)51 6098767; E-mail: antonio.parretta@bologna.enea.it
ABSTRACT
A fluxmeter for parabolic trough solar concentrators is described. The concentrated solar radiation impinging on the receiver of the collector is detected around the focal line by eleven concentration cells distributed on the outer surface of a cylindrical sensor head, each one protected by a translucent window. The sensor head is shaped as a collar to be mounted around the glass tube protecting the cylindrical thermal receiver. Photocurrent and temperature of each cell are measured and the electric signals sent to remote instrumentation. A fan on the back of the sensor body improves the cell cooling. Test campaigns carried out at a solar plant of ENEA-Casaccia (Rome) furnished flux density distributions in good agreement with those simulated on the basis of the shape of the collecting mirrors, which was measured by an optical profilometer.
array of photodiodes, and is moved along the receiver axis between two holders to record a two-dimensional flux map. In order to measure the effective absorbed flux from the receiver. The fluxmeter has been realized with two sensor heads, one measuring the total flux incident on the receiver, the other measuring the total flux lost by reflection. In this paper we present a fluxmeter whose geometry is similar to the PARASCAN but which operates in a different way concerning the selection of the flux effectively absorbed by the receiver.
BASIC SCHEME OF THE FLUXMETER

Fig. 1 shows the basic scheme of the fluxmeter sensor head.
x 6 9 7
INTRODUCTION
The widespread use of solar concentration for the photovoltaic or thermal solar energy conversion demands the development of new instrumentation for measurement of total flux or flux density distribution of the concentrated beam near the receiver. Solar radiation on the ground is 2 typically concentrated at 10-500 suns (1-50 W/cm ) in photovoltaic applications and at thousands of suns (>100 2 W/cm ) in thermal applications. Besides to be suitable to sustain so high flux densities, these fluxmeters are also to be designed to match well the receiver, whose geometry changes with dimension of concentration (2D or 3D) and type of application. For planar receivers, typical of photovoltaic applications, the camera-target method [1] allows to determine in a simple way the flux density profile on a beam section by the image produced on a Lambertian diffuser and recorded by a nearby CCD camera. Absolute flux measurements are also possible by using the fluxmeters developed by Ferriere [2] and Parretta [3]. The matching of a fluxmeter to the cylindrical receiver of a linear concentrator for thermal applications is a more arduous task. Riffelmann [4] has developed a fluxmeter (PARASCAN) whose sensor head is provided with an
y 4 3
z 2 8
Fig. 1. Orthogonal section of the fluxmeter sensor head, with seven cells and large view screens.
The sensor head of the fluxmeter, due to its geometry and the use of concentrating cells as photodetectors, is also called photovoltaic collar (Collare Fotovoltaico, CFV). A photovoltaic collar provided with only seven cells for simplicity, is illustrated in Fig. 1. It is
119
shaped as a polygon (4) on which the concentrating cells (5) are mounted. The sensor head is mounted around the glass tube (3) protecting the cylindrical thermal receiver (2). The cells temperature is measured by thermocouples (8) placed under the cells. The measurement of photocurrent (not shown), coupled to that of temperature, allows to derive the flux density on each solar cell. Finally, the total flux flowing through the cylindrical surface of the CFV is achieved by interpolating the measurements and integrating on 2. From this, the optical efficiency of light collection can be estimated. In practice, the measurement of the effective radiation impinging on the receiver is more complex. Fig. 2a shows, for example, that radiation (6A) is measured by the cell without being collected by the receiver (presence of misalignments), and vice versa that radiation (6B) is collected by the receiver without being measured by the cell (incomplete coverage by cells of the irradiation arc). The first problem is partially solved by the use of screens (9) which select the incoming radiation (see Fig. 2b, c). They are characterized by aperture (acceptance) angle , and are of the large view type (Fig. 2b) or narrow view type (Fig. 2c). The narrow view type screens assure a better selection of light, but are longer than the large view type screens. In Fig. 2 are evidenced the edges (17) and (18), which define the screen aperture. The screens should be blackened in order to avoid unwanted reflections interfering with the direct irradiation. 6A
(measured around z axis), and the total flux incident on a section of the receiver, with thickness z and centered
on coordinate z, can be calculated as follows (see Fig. 3).
C D O z 5 B r1 r2 A 2 z
Fig. 3. Scheme of a beam incident on the cell (5), intercepted by the receiver (2) at point B.
When the receiver is aligned with both light collector and the sun, the solar radiation is focused on z axis. With a calibrated sensor head, the photocurrent density of i-th cell, J iph (i=1, N), is a known function of the flux density,
Ei(c ) , measured orthogonally to the average direction ()

of incident flux:
5 2
J iph ( ) = k ( ) Ei( c )
(1)
a)
6B
9 17 18 b) 5
where () is a function defined in [1], equal to cos() for a Lambertian diffuser. The constant k (in A/W) is the ratio between measured photocurrent density (in A/cm2) and 2 flux density incident orthogonally on the cell (in W/cm ). If r1 is the receiver radius and r2 the distance from the optical axis to the center of the cell (see Fig. 3), then the flux density at the point (I , z) of the receiver is:
Ei( r ) [ z ( )] = cos Ei( c ) [ z ( r2 r1 ) tg ] ( r2 / r1 ) (2)

The concentration ratio (r2 / r1) between flux density on the receiver (2) and flux density on the cell (5) remains unchanged with angle , as such is the ratio C A / B A (see Fig. 3). The average total flux tot (in W) incident on the z-thick section of the receiver, centered on coordinate z, is obtained by integrating the flux density Ei( r ) over the entire arc tot of the concentrated irradiation:
17
c) 18
Fig. 2. a) Cell without screen; b) cell with large view screen; c) cell with narrow view screen.
Fig. 1 shows an example of sensor head equipped with large view screens, where each screen is shared between two adjacent cells. 2 The flux density (W/m ) at a generic point B of the receiver surface, characterized by coordinate z and angle
... = tot r1 z i Ei( r ) / N

with
tot [ z , z ] = tot r1 z E ( r ) ( z ) = ...
(3)
tot in
radians. The same considerations are valid if
we consider a window instead of a cell. We will see in next paragraph, in fact, that the fluxmeter can operate also with cells not directly exposed to the concentrated radiation,
120
but covered by diffusive windows. This allows to reduce cells temperature and to have a more uniform flux on the cells.
THE FLUXMETER
In the following we describe one prototype of CFV realized at ENEA-Portici and tested on a parabolic trough solar concentrator (PCS) located in ENEA-Casaccia (Rome) [5]. The prototype, CFV2, manufactured by CN di Claudio Nappo (NA, Italy), operates with 11 SunPower HECO252 solar concentration cells placed behind translucent targets with quasi-Lambertian transmission properties. This first prototype is devoid of screens which will be applied in the next future; despite this, the fluxmeter has shown to be very accurate. The 11 cells are distributed over an angular extension of ~210 with an angular resolution of ~20. The diffusive input windows are able to intercept around half of the radiation incident on a transversal section of the receiver. An electric fan, placed on the back of the sensor body, improves the cell cooling. Signals of photocurrent and temperature of the single solar cells are measured by remote instrumentation. The monitoring of both temperature and photocurrent of the cells allowed to calculate the incident flux density, after calibration of the sensor at the PASAN mod. 3B pulsed solar simulator, equipped with Fresnel lens for concentration measurements [3]. A schematic view of the CFV2 sensor head is shown in Fig. 4. Each cell (5) is 71.5 mm far from the z axis and 36.5 mm from the receiver (2) surface.
two separable parts, one for mounting the cells, the other (not shown) for accomodate the fan. The sensor is locked on the glass tube (3) through Teflon feet (16). Fig. 5 shows one input window of the sensor head. The concentrated light impinges on the diffusive translucent window (14), is reflected by the walls of the prismatic optical guide (15) and is absorbed by the photodetector (5). This arrangement allows to have on the cell (5) a homogeneous radiation, independent of the impinging direction on the input window (14). The cell photocurrent is obtained by measuring the voltage drop on a nearby 0.01 shunt resistance. The cooling of cells is performed by circulating air in the space (16) existing between the top cover (10)+(11) and the base (4) of the sensor head.
15 14
10
15 14 11 11
8 8 13 12 13 12 5 5 4 4
10
16 16
Fig. 5. Section of a sensor head window.

The sensor CFV2 is protected by eleven aluminium screens (10) coupled to teflon screens (11), to assure a good thermal insulation for the base (4) of the sensor where are accommodated the cells (5). The diffusive window (14) assures a flux on the cell proportional to that impinging on the window, but independent by its spatial and angular distribution. The angle-resolved response of the window/cell system is being studied at the PASAN pulsed solar simulator. Photos of the sensor head assembled and mounted on a glass tube simulating the true glass lining of the solar receiver are shown in Fig. 6. On the right side the electric fan improving the solar cells cooling is visible.
14
10
15
y
16
Fig. 4. CFV2 sensor head mounted on the glass tube (3) (125 mm outer diameter) of the PCS concentrator.
The arc covered by the set of eleven cells is 216, then the distance between adjacent cells is 21.6. The single cells subtend central angles ~ 10. The sensor has been realized from a cylindrical body of aluminium (4), divided in
Fig. 6. Top of the CFV2 sensor head with visible the diffusive windows (14) (left); back of the sensor head with visible the electric fan (right).
The test campaigns carried out at the solar plant of ENEA-Casaccia (see Fig. 7) proved the suitability of the sensor to sustain the radiation flux densities (around 80 suns) there produced. Tests at high irradiation showed that the cells temperature ranges from 60 to 80 C, well
121
below the maximum temperature recommended by the manufacturer (100 C).
an example of outdoor measurement: on the top is the picture of the sensor head taken by the vertex of the parabolic collector; on the bottom, the flux measured by the CFV2 sensor, compared with the simulations for real (AC01) and ideal (parabolic shape) mirrors. The shadowing due to the receiver is clearly shown in the picture; the other dark regions are due to the modulation of flux distribution shaped as a reverse (bottom). The agreement between measured and simulated flux worsens at high values, where the response of windows (14) ceases to be Lambertian, then a suitable correction has to be applied to the measured data.
CONCLUSIONS
A method for high flux density measurements in trough solar concentrators has been discussed. The concentrated radiation is measured by eleven SunPower HECO252 concentration cells distributed over the outer surface of a cylindrical sensor head, and protected by diffusive windows with quasi-Lambertian transmission properties. The device has shown to be a valid instrument for the accuracy shown with outdoor tests on a solar trough concentrator operating at around 80x suns.
Fig. 7. Photo of the 11-cells sensor head mounted on the PCS receiver during a test campaign.
REFERENCES
[1] A. Parretta, C. Privato, G. Nenna, A. Antonini, M. Stefancich. Monitoring of concentrated radiation beam for photovoltaic and thermal solar energy conversion applications. Applied Optics 2006. 45, pp. 7885-7897. [2] A. Ferriere, and B. Rivoire. An Instrument for Measuring Concentrated Solar Radiation: A Photo-sensor Interfaced with an Integrating Sphere. Solar Energy 2002. 72, 187-193. [3] A. Parretta, A. Antonini, M. Armani, G. Nenna, G. Flaminio, M. Pellegrino. Double-Cavity Radiometer for High Flux Density Solar Radiation Measurements. Applied Optics, in press (20 April 2007).
Flux (1/rad)
50
Signal / Rad_eff (arb. u.)
40 30 20
10:00 azi = -32.0 deg CFV2 ideal AC01
1,5
1,0
0,5 10 0 -100 -50 0 (deg) 50 100 0,0
[4] K. J. Riffelmann, A. Neumann and M. Wittkowski. Parascan: a new parabolic trough flux scanner. ISES Solar Worl Congress, Gteborg, June 2003, ISBN 91-6314740-8. [5] A. Parretta, C. Privato, L. Gentilin, A. Maccari, A. Mittiga, M. Montecchi, M. Tucci, G. Nenna, Radiometro per ricevitori cilindrici, Brevetto It., Application N. BO2006A000880, 27 Dicembre 2006. [6] A. Maccari, M. Montecchi. An optical profilometer for the characterisation of parabolic trough solar concentrators. Solar Energy 2007. 81, pp. 185-194.
Fig. 8. (top) The sensor is viewed from the vertex of the parabolic mirror; (bottom) experimental flux (CFV2) compared with simulations for real (AC01) and ideal (parabolic shape) mirrors.
The test measurements furnished the flux density near the focal line of the solar collector. The distributions were in good agreement with simulations based on experimental data of the optical profile of the collecting mirrors, as obtained by indoor laser measurements [6]. Fig. 8 shows
122
Inverters response time with concentration PV systems

S.Chellini, R.Pardell Sol3g, S.L., Parc Tecnologic del Valls, 08290 Cerdanyola, Barcelona, Spain
ABSTRACT After our field experience, based on data collected in different installed systems, MPP control inverters show a long time response after a drop in irradiance. This effect, that may not affect flat panel based systems significantly, have a big impact on concentration PV systems energy collection efficiency during days of intermittent clouds and sun spells, when light losses its direct irradiancy component temporally. Concentration modules fall in these short periods into nearly zero power production that implies that the inverters fall into stand-by mode. When the DNI eventually comes back to usable levels, the MPP control restarts calculating the optimal combination of voltage and current from scratch, operation that requires a not negligible time. New MPP tracking algorithms which remember previous MPP voltage levels must be implemented in inverters in order to improve energy collection efficiency for concentration PV systems. INTRODUCTION Sol3g has installed and tested to date three different triple junction based PV high concentration pilot systems. The systems range from 760 to 920 W power at the DC side under working conditions, rated at 1.000 W/m2 DNI. Data used in this paper comes from a system located in Mont-Ras, a locality close to Girona, Spain, positioned in the geographical site: long 41 55N and lat 3 13E. Its power is rated at 760 W. All three pilot systems use 28 HCPV modules designed and manufactured by Sol3g in its PTV facilities at Cerdanyola del Valls. The module is constituted by a primary optic (a ten Fresnel lens parquet 1,20 meters long), a secondary optic (a Borosilicate prism) and Triple Junction photovoltaic cells of 5.5x5.5 mm size. Its characteristics are: 22.7% module efficiency (@ 1000 W/m2 and 25 C) [1] 30 V open circuit voltage. 1,2 A short circuit current. The tracking system used is made by Feina, a company located in Catalunya, Spain, with whom we cooperate. The tracking system follows the sun following an hybrid strategy: using an attached external sensor in closed control loop when the sun is visible, and calculating
the theoretical position in open control loop, by means of an astronomic algorithm, when the sky is covered.
Fig. 1: An image of the site installation at Mont-Ras The modules are connected in two grids of 14 modules each, containing two series connected groups of 7 modules, and then parallel connected. The DC output passes through an electronic device, which sends the signals to the data logging system. We use to acquire the I-V two analogical/digital converters, I-7018 ICP Con, able to read eight different DC signals in the range -100 mV; +100 mV. The electronic device modifies voltage PV DC output from 0-300 V to the appropriate range; the PV DC current signal has to be modified into a voltage signal, by a shunt resistance. We also connect to the A/D converter different sensors: Two temperature sensors (K-type thermocouple): one to measure the ambient temperature and another for the module temperature. A humidity sensor. Three calibrated silicon (Spektron) cells to measure: the direct normal irradiance (DNI), adjusting a matt black tube on the top of the cell, to collect only the direct component of the light; the global normal irradiance (GNI), attaching the cell to the tracking system; and the global horizontal irradiance (GHI). The data passes through a client box able to communicate to the PC by means of RS-485 protocol, and the values are collected every day in a different file, by software created by Sol3gs (SOLMON program).
123
Day 03/12/2006
singular point
20,0% 900
eta DNI GNI power
800
700
15,0%
600 DNI - GNI - Power
500 (%)
10,0%
400
300
5,0%
200
Fig. 2: Data logging system scheme

0,0% 08:52 10:04 11:16 12:28
shades
100
13:40
14:52
16:04
0 17:16
The system is connected to the String Inverter SWR 700 (SMA), a device with maximum power point seeking integrated in its functionalities. One important thing to underline, is that the operation state is reached when the string voltage (UPV) at the inverter is minor than 50 V. This can be very important for the concentration photovoltaic systems, because in days with low light irradiance and in absence of the direct component of light, voltage could drop to a value below 50 V, which permits to enter in the stand-by state. The tracking system follows moving every 30 seconds to the theoretical sun position, so, during a day of intermittent clouds and sun spells, the concentration system is able in each favourable break to inject power to the net, but the inverter has to re-calculate the maximum power point, scanning the entire I-V curve and researching the optimal forward bias.
Time (hh:mm)
Fig. 3: Day 03/12/2006, sunny day with a maximum Tmod 35 C; transit at 12:36. The graphic in Fig. 3 represents time dependence of the direct normal irradiance (DNI), global normal irradiance (GNI), the generated power of the installation and finally the conversion efficiency (). The generated power reaches a maximum only closed to the sun transit, because of shades those partially covers the tracking system during the morning.
Day 30/01/2007
20,0% 18,0%
singular point
eta DNI GNI power 1000

singular point
16,0% 14,0% 12,0%
800
(%)
600 10,0% 8,0% 400
EXPERIMENTAL RESULT
We will show some data collected during the period September 2006-February 2007, representing the DNI (W/m2), GNI (W/m2), the generated power by the system, P (W) and its conversion efficiency (PV performance [2]) calculated as:
6,0% 4,0% 2,0% 0,0% 08:52 0 17:16
200
10:04
11:16
12:28
13:40
14:52
16:04
Time (hh:mm)
P = max 100 Etot A
Eq. 1
Fig. 4: Day 30/01/2007, cloudy day with a maximum Tmod 32 C; transit at 13:00. The graphic in Fig. 4 represents the time dependence of DNI, GNI, P and in a cloudy day, with a very variable direct irradiance in the beginning of the day and after two PM. A detailed analysis of the DNI and power values (Fig. 5) in the 14:49 14:58 interval shows two different power response delays: at 14:50 DNI raises 2 to 500 W/m for 1 minute period, but power remain at the 2 zero level; at 14:56 the DNI reaches 500 W/m and raises up till 700 W/m2, and the power takes 1.5 minute long to raise to the normal production (in those ambient condition was 450 W).
where Pmax is the instantaneous power given by the system, Etot is in this case, the direct component of the light (DNI) and A represents the input aperture surface of 2 the system (in our case 4.032 m ). The SOLMON program has showed some incorrect values of DNI and GNI (and indirectly of the efficiency), caused by punctual incorrect readings of the I-7018 A/D converter (singular points in the graphic below).
124
DNI (W/m2), GNI (W/m2), P (W)
18,0% eta 16,0% DNI GNI 14,0% power 750 850
650
550 10,0%
(%)
8,0%
450
6,0%
350
DNI (W/m2), GNI (W/m2), P (W)
12,0%
The inverter expends a lot of time trying to find its new optimal voltage level, and we see that instead of starting looking at this point from a zero voltage, and then going up to an excessive voltage, it would be much more efficient to start from the last MPP voltage reached before the DNI fell near to zero. In fact we also see that under very unstable irradiance conditions the MPP tracking algorithm is somewhat unaccurate, as it is making the PV generator work at a voltage around 160 V, which is in fact not the average MPP voltage of 180V, thus affecting the efficiency of the system during the very unstable DNI episodes. As can be seen, the combined effect of both problems (high response time and MPP lack of accuracy) on integrated energy collection efficiency can be very severe during days where the DNI shows a binary behaviour.
Day 04/11/2006
3 560 2,5 480 DNI GNI voltage current
4,0%
response delay
250
2,0%
150
0,0%
50
-2,0% 14:49
-50 14:51 14:52 14:53 14:54 14:55 14:56 14:57 14:59 Time (hh:mm)
Fig. 5: Day 30/01/2007, detail of an irradiance drop. The GNI during this period never falls below 200 2 W/m , which permits to a flat panel to generate power during all the period. Detailed analysis of many similar episodes and of the integrated daily efficiency under different direct irradiance conditions show that the slow response of MPP tracking inverter reduces the overall energy collection efficiency of the system.
DNI (W/m2)- GNI (W/m2) - Voltage (V)
DISCUSSION
An I V analysis shows us how voltage and current behave during a day: Fig. 6 represents a sunny day (03/12/2007, reported before), with no great temperature variability (the ambient temperature flattens at 20C during the day, from 10:00 to 17:00, and the module temperature reaches 34 C at 14:00, but it is steady in the rest of the day at 30 C). The result is that the current follows the DNI shape, but the voltage remains approximately constant to 180 V.
Day 03/12/2006
240 3 voltage current
400 mpm search point
320
1,5 240 1 160 0,5 80

response delay
0 12:00
0 12:02 12:05 12:08 12:11 12:14 12:17 12:20 12:23 Time (hh:mm:ss)
Fig. 7: Day 04/11/2006, detail of variably DNI component (notice the stability of the GNI). The graph shows time dependency of Voltage and current in the selected range.
CONCLUSION
In this paper, different graphics are shown, using data collected from a high concentration PV system working under different irradiance conditions. We notice a long response delay between the DNI fluctuations and the MPP inverter control when the system has to work under binary variation of direct irradiance, typical of days in which cloudy conditions alternate with sun spells. The more variable the irradiance conditions the greater the effect this slow inverter response will have on the integrated energy collection efficiency of the concentration PV system. In the other hand, we propose to add a voltage control to the inverter electronic device, able to remember the operative Vmpm voltage of the system in order to reduce its response time.
210 2,5 180 2 Current (A) 150 Voltage (V)
120
1,5
90 1 60 0,5 30
shades
0 9:50:24 11:16:48 12:43:12 Time (hh:mm:ss) 14:09:36 15:36:00 17:02:24
0 8:24:00
Fig. 6: Day 03/12/2006, time dependence of Voltage and Current for the tracking system. We can see in Fig. 7 I V values of another cloudy day (04/11/2007), in which the system disconnects to the inverter when the DNI falls nearly to zero. From 12:15 the DNI starts to raise, and the MPP control inverter searches for the MPP, and the system works initially at more than 200 V, and it finds the maximum power point voltage at 160 V, which is close to the last Vmpm found at 12:05.
125
Current (A)
As can be gathered from the data, the systems voltage is quite stable around 180 V whenever the inverter reaches MPP. Therefore, a very simple modification of the inverter logic, making its software remember the latest MPP reached voltage, would help drastically reduce the response time from more than one minute to a few seconds.
REFERENCES
[2] F.Chenlo, J.P.Silva Informe tecnico modulos Sol3g 2006. Ciemat, Madrid. [2] A.Luque, S.Hegedus Handbook of Photovoltaic Science and Engineering 2002. 16.1 pp. 702-715.
126
PREDICTION OF PV CONCENTRATORS ENERGY PRODUCTION: INFLUENCE OF WIND IN THE COOLING MECHANISMS. FIRST STEPS
M. Martnez, I. Antn, G. Sala Instituto de Energa Solar, Universidad Politcnica de Madrid (IES-UPM) E.T.S.I. Telecomunicacin, Ciudad Universitaria s/n, 28040, Madrid, Spain Tel:+34915441060, Fax: +34915446341, mail: maria.martinez@ies-def.upm.es
ABSTRACT
Making a complete model of a PV Concentrator (CPV) is essential to obtain a good prediction of its energy production. For that, it is necessary not only to define the electrical characteristics of the system in standard test conditions (STC), but also to determine the influence of the ambient conditions on the electrical performance. These conditions are related with the cooling of the system, moreover, a proper cooling of a CPV is essential to obtain a good performance in operation. This work is going to be focused in only one type of CPVs, the Point Focus CPVs based in lens optics, because they have a particular characteristic: its classic housing and the internal air are going to contribute to the system cooling. We have find out that an important amount of heat is going to be dissipated through this region and not through the module back heat-sink. We have carried out the first steps of this work, obtaining a thermal behaviour model for this type of CPVs in steady state and calm air. We have used an experimental method, based on temperature measurements and incoming solar power cast on the system. So, finally, we are able to of identify all parameters related with the cooling process and finding out their relation with ambient conditions.
reach very high values if the cooling is not the appropriated, because, in this kind of systems, the incoming radiation is multiplied by the effective concentration factor (C), so all the light that is not converted into electrical power has to be efficiently evacuated of the system. This is why CPVs have a very complex cooling and, when passive cooling is used, wind is the agent that has more influence in this mechanism. In this work, we are going to center our efforts in characterizing only one kind of CPVs: the Point FocusLens housing CPVs. This type of concentrators has a special characteristic that is related with its cooling, they are closed volumes with an internal interface of air that is going to be involved in the cooling mechanisms of the system.
OBJECTIVES
The main objective of this paper is to describe the existing connection between the operating cell temperature and the ambient conditions. Once we have obtained this relation, the change of the energy production rate to other operating conditions will be simply calculated, what will make the CPV energy prediction easier. To reach this objective, it is necessary first, to study the thermal behavior of this type of CPVs in steady state, defining all the parameters involved in the cooling mechanisms and identifying which are going to be influenced by wind.
INTRODUCTION
An accurate modeling of PV Concentrators (CPV) for the prediction of its energy production is essential to establish the economic viability of this kind of systems at different sites. For that, it is necessary not only to define the electrical performance of the system at standard test conditions (STC) but also to determine the dependence of its electrical behaviour on the ambient conditions. The output power of this kind of systems depends on the direct irradiance level, light spectrum, ambient temperature and wind direction and speed. This last variable, wind, is the most difficult to model and this work focuses on the correlation between the power and the wind in a CPV. In a CPV, the evacuation of heat is an essential and complex task [1]. The operating cell temperature could
APPROACH
In this section, we attempt to enumerate and describe the critical points of the CPV energy prediction process. CPV characterization For characterizing a CPV in STC we have first to measure the DC power of the system. From its IV-Curve 1 1 (I1, V1), in whatever conditions of operation (B , Tcel ) and making a transfer to the STC (B*, Tcel*), with equations defined below (1), (2) (single junction cells), we obtain the IV-Curve (I2, V2) in STC.
127
Table 1 collects the definition of certain parameters of the concentrator and constants which will be necessary for the calculation.
I 2 = I1 B* B1
Radiation
q = T24 T14 A
(6)
Emissivity
(1)
Stefan-Boltzmann constant
q = hrad T A = (T2 + T1 ) (T22 + T12 )
E T* V2 = V1 + N S g (V1 + I1 RS ) 1 cell + 1 e Tcell NS

* T * B* k Tcell B* Ln 1 m Ln cell I1 1 1 RS B B T1 e cell
(7) (8)
(2)
hrad
Table 1. Characteristic parameters of a PV concentrator and essential constants

Characteristic parameters of the system NS Number of cells connected in series in the system Eg Cell band gap RS Series resistance of the system, cells and wires m Diode quality factor Constants k e Index which fluctuates between 2 and 4 Boltzmann constant Electron charge
The general radiation equation, (6), can be written like equation (7) via the radiation coefficient defined in equation (8).
THERMAL BEHAVIOUR
Description of the problem As we said at the beginning of the paper, it is necessary to make a thermal study of the system, more in detail; we want to know what are the heat flowing ways from the cell to the surrounding atmosphere [1]. To reach this purpose, we present in Fig. 1 a general thermal scheme, useful for this type of CPVs.
Lens Ta P5 P6 Tai T cel Plight P2 P4 P6
Energy production prediction The performance of the system at any time is defined also by equations (1) and (2) as a direct function of irradiance and cell temperature. The influence of the ambient conditions in the energy production its captured in the parameter operating cell temperature (Tcel). This parameter can be calculated with equation (3) taking the ambient temperature, direct beam radiation and power of the system as inputs.
Tcel = Ta + (Plight Pelec ) Rth , sys
(3)
P1 Tplate Back Al plate P3 Housing walls
It is here in equation (3) where the parameter thermal resistance (Rth,sys) appears. This parameter is directly related with the thermal behaviour and cooling of a CPV and will be one of the results obtained from this work. Cooling mechanisms in CPVs For passive cooling the main mechanisms involved are [2]: Conduction: For carrying the heat from the cell to the dissipating surfaces.
q T = k x A
Ta
Fig. 1 Scheme of a Point Focus-Lens Housing CPV In this figure, we can see all the possibilities that have the heat for flowing from the cell to the ambient air. In addition, as we said before, we have to keep in mind the internal interface of air, taking into account the heat dissipation through the lateral housing walls and the front lenses. The thermal behaviour of a CPV can be represented as an equivalent thermal circuit, Fig.2. The heat flow and the temperature perform in the same way the current and the voltage do in electric circuits. Thermal resistors substitute ohmic resistors. From equations (4), (5) and (7), we can define the thermal resistances for all mechanisms involved in the CPV cooling.
Rth ,cond = T 1 = x P kA
(4)
k Thermal conductivity
Natural Convection and Radiation: For removing the heat from the dissipating surfaces to the surrounding atmosphere.
q = hconv T A
Natural Convection (5)
h Convection coefficient
(9)
128
Rth ,conv = Rth , rad
T 1 = P hconv A T 1 = = P hrad A
(10) (11)
Because both, convection and radiation, are the heat removal mechanisms from the dissipating surfaces, it is useful to define a combined convection and radiation thermal resistance.
Rth ,comb =
Because the main purpose of this work was to describe the wind influence in the energy production of this type of CPVs, then the first task is to identify which parts of the system are sensitive to wind. It is obvious that the surfaces in contact with the surrounding atmosphere are going to be the ones influenced by wind. In this preliminary work, we are only going to describe its behaviour in calm air; the influence of wind on these parameters will be analyzed in future works. In the equivalent thermal circuit the action of wind is simulated by variable convection resistances. Solving the problem Some simplifications in the circuit will help to obtain conclusions. As we said before removing heat from any surface is made by convection and radiation, but for bright metallic surfaces the convection coefficient (hconv) is much bigger than the radiation coefficient (hrad), so we do not have to take into account the irradiation mechanisms in the heat flow dissipating from the metallic surfaces of the CPV.
(hconv + hrad ) A
Rconv Rrad Rconv + Rrad
(12)
Tcel P2
Rcond,cell-plate R cond,plate Tplate P3 P1 P4 Rcomb, cell Rcomb, plate Tai P5 P6 Rconv,wall Rcomb, h-s R cond,len
Pheat Rconv , len
Rrad , plate Rrad ,h s Rrad , wall
(13)
Rcond,wall
R comb, len
Rcomb, wall
Another aspect to take into account is the heat directly flowing from the cell to the internal air. It can also be rejected, because the available area for dissipating heat from the cell (Acel) is much smaller than the back Al plate area (Aplate) which uses both sides for heat dissipation, to closed internal and open external air.
Rcomb ,cell
(14)
Ta
Fig. 2 Equivalent thermal circuit for a Point Focus-Lens Housing CPV Fig.2 shows the equivalent thermal circuit corresponding to the thermal behaviour of this type of CPVs represented in Fig.1. All the surfaces and interfaces involved in the system cooling are represented by their equivalent thermal resistance; Table 2 explains the meaning of each resistance. Table 2 Interfaces involved in the cooling of a Point FocusLens Housing CPV
THERMAL RESISTANCES Between the cell and the back Al plate cell-plate Through the back Al plate plate Between the cell and the internal air cell Between the back plate and the internal air plate Between the internal air, through the housing wall walls and to the ambient air Between the internal air, through the lenses and to len the ambient air Between the heat-sink and the ambient air h-s
The biggest approximation we made for solving the problem was to consider an equivalent thermal resistance (Req) that represents the capacity of dissipating heat from the internal interface of air to the surrounding atmosphere. This approximation opens a future work for studying more in detail all the parts involved in this heat dissipation.
Req =
(R (R
+ Rcond ,len + Rcomb ,len ) (Rconv , wall + Rcond , wall + Rcomb , wall ) + Rcond ,len + Rcomb ,len ) + (Rconv , wall + Rcond , wall + Rcomb , wall ) conv ,len
conv ,len
(15)
Finally, for making easier the tackling of the problem we suppose isothermal surfaces for all heat exchanging surfaces.
Rcond , plate 0
(16)
Fig.3 shows the simplified circuit we have used to solve the problem of the thermal behaviour of a Point Focus-Lens Housing CPV in steady state and in calm air. The general equations associated to this circuit and needed for solving the problem are:
Pheat = Plight Pelec
Pheat = P = P3 + P4 1 Tcel T plate = Rcond ,cell plate P 1
(17) (18) (19)
129
Tplate Tai = Rconv, plate P4

T plate Ta = Rconv , h s P3
Tai Ta = Req P4
(20) (21) (22)
CONCLUSIONS
The initial purpose of this work was finding the relation between the cooling effect of module and wind direction on the energy production of Point Focus-Lens housing CPVs. Only the first steps are presented here, by the time of ICSC-4 Conference. This paper only gathers the description of the thermal behaviour in calm air in steady state. The use of equivalent thermal circuit models has allowed to identify the main flows of heat and the key measurements of temperature an power input that led to useful results, like the identification of the overall thermal resistance between cell and air, (Rth,sys). The next steps identified for progressing in this work are: a) The study of the heat flow through the lateral walls and the front lenses, eliminating in future experiments several unwanted flows during test. b) The effect of the array position on the natural heat convection must be analyzed. c) Finally, the final task will consist to measure the wind effect in the cell cooling mechanisms. We have already carried out several experiments in this way, blowing forced wind on our array sample and obtaining experimental data of temperature and power variation, but we dont have yet consistent results.
Rcond,cell-plate Tcel P Tplate 1 P4 Rconv, plate Pheat
P3
Tai R eq
Rconv, h-s
Ta
Fig. 3 Simplified equivalent thermal circuit for a Point Focus-Lens Housing CPV
RESULTS
We solved the problem related with the simplified circuit represented in Fig.3 applying equations (17) to (22) for one type of these CPVs like a case of study. The preliminary results obtained are: First of all, that an important amount of heat is dissipated through the internal interface of air. For this case of study, we obtained that 23% heat flow is not dissipated through the system heat-sink back side. Therefore, this heat flow from the front lenses and side walls must be taken into account in future works with this type of CPVs. In addition, we have obtained a relation for the temperature distribution through the parts of the system in this case of study. For an overall 50C temperature drop between the ambient air and the cell, there are 20C of difference between the cell and the back Al plate. Also, we can advance that the internal air, is about 20C hotter than the ambient air. The main result is that we have modeled the thermal behaviour of this type of CPVs in steady state and in calm air. Experimentally we have obtained the value of the thermal resistance of the system (Rth,sys) for the case in study. Moreover, with equations (3) and (23), we have directly obtained a method for finding out the operating cell temperature taking the ambient temperature, direct beam radiation and power of the system as inputs.
Rth , sys = Rcond ,cell plate +
ACKNOWLEDGEMENTS
We want to thank for the manufacturer that has lent us its prototype for carrying out our study.
REFERENCES
[1] G. Sala. Cooling of solar cells, Cells and optics for Photovoltaic Concentration, Chap.8, pp. 239-267, Adam Hilger, Bristol, 1989. [2] A. J. Chapman. Transmisin de Calor Bellissco, 1984.
(R (R
conv , plate
conv , plate
+ Req ) + Rconv ,h s
+ Req ) Rconv , h s
(23)
130
FLUXMETER FOR POINT-FOCUS SOLAR CONCENTRATORS

A. Parretta*1, A. Antonini2, M. Stefancich2, G. Martinelli2, M. Armani3 ENEA Centro Ricerche E. Clementel, Via Martiri di Monte Sole 4, 40129 Bologna (BO), Italy. 2 Physics Dept., University of Ferrara, Via Saragat 1, 44100 Ferrara (FE), Italy. 3 Institute for Renewable Energy, EURAC reasearch, Viale Druso 1, 39100 Bolzano (BZ), Italy. *Phone: +39 (0)51 6098617; Fax: +39 (0)51 6098767; E-mail: antonio.parretta@bologna.enea.it
1
ABSTRACT
A fluxmeter for high flux density measurements in point-focus solar concentrators is based on the use of two integrating spheres, coupled by an intermediate window of selected aperture area. The concentrated radiation is collected by the first sphere through an input window, integrated and driven to the second sphere where it is coupled to a conventional radiometer and to a spectrometer for flux and spectral measurements, respectively. The overall attenuation factor of input radiation is controlled by selecting the area of intermediate window. Attenuation levels from few tens up to few thousands can be well controlled. The fluxmeter has been calibrated by a pulsed solar simulator modified to operate with concentrated radiation at levels of hundreds of suns. An optical model and a ray-tracing study have been also developed and validated, by which the optical and thermal properties of the fluxmeter have been fully explored.
operate as integrating spheres, after deposition of a white diffusive coating. The concentrated radiation (cl) enters the first cavity (is1) through the window (win), then enters the second cavity (is2) through the intermediate window (wco), whose aperture is varied according to the intensity of input beam by using inserts provided with holes of different area. Cavity (is2) has a photodetector (pd) for flux measurement by radiometer (rad), and an optical fiber (of) for spectral measurement by spectrometer (sp). The input window (win) is selected depending on the type of measurement being carried out. It provides the input of the focused beam for total flux measurement, or a portion of the beam for flux density mapping. In the last case the radiometer can be moved in front of the beam by an electronically driven x-y translation stage.
cl win is1 wco wof is2 wpd pr pd of
sp
INTRODUCTION
The characterization of the concentrated solar beam is a direct way to perform the optical characterization of a solar concentrator. Among the direct methods, one of the most used is the camera-target one [1, 2], which gives the irradiance profile of the beam on the surface of a Lambertian target. We have theoretically analyzed the camera-target method for planar targets and produced an associated image reconstruction algorithm [3]. For a complete optical characterization of the beam, however, supplementary measurements of the absolute flux density are required, at least on some points of the tested plane, by reference calorimeters [4]. In this paper we present the last results obtained by developing a fluxmeter devoted to point-focus solar concentrators of the Fresnel type, also useful for indoor testing of small concentrator prototypes, easy to operate and portable. The fluxmeter operates under a stationary irradiation regime and the photodetector is thermoregulated to avoid variation of sensitivity at changing irradiation level. The fluxmeter provides also the spectral irradiance distribution of the concentrated radiation.
rad Fig. 1. Scheme of the fluxmeter. THE OPTICAL MODEL

The fluxmeter has been optically modelled by applying the conservation law for the input flux. The main parameters of the model are: the flux 0 at input window (win) of area Sin; the diameter d and the wall reflectance Rw common to both spheres; the aperture area Sco of the intermediate window (wco); the flux m on the photodetector (pd) with area Sm and reflectance Rm; the area Sof and reflectance Rof of the optical fibre (of) head; the irradiance G0 at input window (win), the irradiance G1 on the wall of cavity (is1) and the irradiance G2 on the wall of cavity (is2). We define also the attenuation factor for flux, f A , as the ratio between flux 0 at input and flux m on the photodetector (pd):
THE METHOD
Fig. 1 shows the basic scheme of the fluxmeter. The two interconnected optical cavities (is1) and (is2) are obtained by hollowing an Aluminium prism (pr) and
131
f A = 0 / m
(1)
By applying the flux conservation law, we obtain for the attenuation factor:
f A = Sin G0 / Sm G2 =
1 2 Sco Sm Sco
(1)
where:
= [ S in + S co + S w 1 (1 R w )] ;
= [ S m (1 R m ) + S of (1 R of ) + ...
Fig. 3. Intensity of irradiation in the two cavities of the fluxmeter after a ray tracing study.
A simulated light beam, of known power, number of rays and angular distribution, enters the first cavity, having the concentration region (CR) exactly correspondent to the entrance aperture (win) of the radiometer. Part of light is lost back through (win) (not shown), part of light is transferred to the second cavity. In fig. 3, at the stationary state, we distinguish the power content of each ray by its colour. Moreover, the distribution of colours in the two cavities allows to qualitatively estimate the irradiation intensity and distribution. It is evident the attenuation the light undergoes moving from the (is1) to (is2), and the more uniformly distributed light intensity in (is2) with respect to (is1). In this last cavity, moreover, is also visible a central region on the bottom, facing the input aperture, that is more illuminated as it corresponds to the area of first impact of the input beam. The ray-tracing model allows to analyze in detail the irradiation distribution into the cavities. It is evident that an additional integrating sphere assures a better distribution of light on the photodetector and then a less dependent response on the input beam geometry. We have analysed, for each cavity, six different portions of the wall, with the same area. For each portion we have calculated the incident flux, normalized to the average value calculated for each sphere. The results are reported in Fig. 4. As can be seen, one decade smaller statistical variations are observable on the flux density in the second sphere with respect to the first one.
1,2
is1; sd=0,295 is2; sd=0,0265
... S co + S w 2 (1 R w )]
and where Sw1 is the wall surface area in sphere (is1) and Sw2 the wall surface area in sphere (is2), different between them as different are the window openings. The optical model has been tested fixing all parameters except the aperture area Sco and the wall reflectance Rw. The fixed 2 parameters were; Sin = Sm = 1.21 cm ; d = 5 cm; Sof = 0.2 2 cm ; Rm = 4%; Rof = 84%. The SunPower HECO252 concentration cell was used as photodetector. Fig. 2 shows the attenuation factor for flux f A , calculated as function of intermediate aperture area Sco from 0.1 to 2.0 2 cm , for a set of wall reflectivities from 92% to 99%. As expected, high attenuation values are obtained at small window aperture and low wall reflectivity. In the first case less power is transferred from the first to the second cavity, whereas in the second case more power is lost as heat on the walls of the two cavities. To expand the dynamics of the attenuation factor, the area Sco is changed by using inserts with different aperture area, which are introduced in a slit between the two spheres. The values of f A shown in Fig. 2 are restricted to a few hundreds. Higher values, in the order of thousands, are obtained by reducing the input window area Sin. A raytracing study on the fluxmeter has been performed by using the TracePro code. The simulation has been applied to the original CAD project, the same used by the manufacturer to realize the fluxmeter prototype. Fig. 3 shows a typical optical simulation.
200
Attenuation factor, fA
Sin = 1.21 cm 150
Flux density
100
Rw92% Rw93% Rw94% Rw95% Rw96% Rw97% Rw98% Rw99%
1,0 0,8 0,6 0,4 0,2 0,0 1 2 3 4 5 6
50
0 0,0
0,5
1,0
1,5
2
2,0
Selected area
Aperture area, Sco (cm )
Fig. 2. Attenuation factor for flux f A .
Fig. 4. Normalized average flux density on six surface portions (4 cm2 area) of the internal wall of each cavity (sd = standard deviation).
132
A uniform irradiation distribution into the second cavity is a prerequisite to obtain a constant response of the radiometer, in terms of attenuation factor, at constant power in input. The second aspect to investigate in the next paragraph is the response of fluxmeter to the angular divergence of the beam into the first cavity at constant flux at input.
THE FLUXMETER
Fig. 5 shows the general scheme of the radiometer, which includes all the accessories and instrumentations needed for thermoregulation of photodetector, for measuring the photocurrent Iph and the spectrum of radiation.
rw rd
cl Pe of
rd ins rd pr
tr
sp
The insert (ins), whose aperture area can be changed between ten and hundred square millimetres, is introduced in the slit opened in the middle of the prism between the two cavities. The spectrograph (sp) is connected to the gauge head by an optical fibre (of), placed just in front of the detector. The complete fluxmeter was fabricated by ECOVIDE (Italy). A photo of the fluxmeter is shown in Fig. 6. The fluxmeter was calibrated by the PASAN mod. 3B pulsed solar simulator. A Fresnel lens was used to concentrate the light of the simulator on a 1 cm2 area. The calibration was made by measuring the photocurrent of a SunPower HECO252 test cell used as reference (1.21 cm2 area), the same used as photodetector in the fluxmeter, and then by measuring the photocurrent given by the fluxmeter. The input window (win) (1.4x1.4 = 1.96 cm2 area) was suitable to accept all the flux of the beam CR. This operation was repeated for different levels of light concentration and for different types of inserts. The experimental values of attenuation factor f A have been
f cm Iph P(heat) T (C) ts
compared to the theoretical data of the optical model and

of raytracing analysis. The comparison is better outlined by a log/log plot and a linear fitting of the data:
2,8 2,6 2,4
fA exp fA opt fA optda fA ray

Figure 5. General scheme of the fluxmeter.

The control module (cm) provides the electric power P(heat) needed to drive the Peltier cooling system (Pe), cooled by the fan (f), and to control the temperature of the SP-HECO252 cell in the 5-60 C interval (typical operational temperature is 25C). The temperature T(C) of the cell is set out through the touchscreen (ts) on the (cm) and is measured by the thermistor (tr). The photocurrent is obtained by measuring the voltage drop on the 0.01 ohm shunt resistance (Rsh). The gauge head of fluxmeter (pr) was fabricated by working a prism of Aluminium, and is made of two sections joined at the horizontal plane. It can be protected on the front side by a reflective wall (rw) acting as a mirror and reducing the heating of accessories. The first cavity, the most involved in the heating process, is cooled by finned radiators (rd) placed on four walls of the prism.
Log fA
2,2 2,0 1,8 1,6 -1,0 -0,8 -0,6 -0,4 Sin=1.96cm

2
-0,2
0,0
Log Sco
Fig. 7. Experimental (exp) and theoretical data of attenuation factor for flux f A . Suffixes: opt = optical model; optda = optical model + dark area; ray = raytracing model.
log f A = a + b log S co
(2)
Fig. 7 shows the linear plots. The data from the ray-tracing model give the best matching with the experimental ones. The optical model has been artificially improved by adding 1 cm2 of black area in each cavity to simulate the light absorption due to regions of the cavities difficult to model. In conclusion, the calibration measurements give for the attenuation factor the relationship:
log f A = 1.66 1.01 log Sco
(3)
Fig. 6. Photo of the fluxmeter gauge head.
We have analysed so far the fluxmeter at a fixed geometrical configuration of the beam and in absence of
133
Relative attenuation factor
baffles. A baffle, however, is necessary in the first cavity (see (b1) in Fig. 8a), when the input beam divergence is high enough that some radiation could directly enter the second cavity from the input window. Similarly, two other baffles are necessary in the second cavity for hiding the photodetector and the optical fiber from light entering (is2) though (wco) (see(b2) and (b3) Fig. 8b).
1,15 1,10 1,05 1,00 0,95 0,90 0,85

Old, no baffles Old, baffles New
win is1
a)
pr z wco is2 pr y x
b1
10
20
30
40
50
60
70
80
90
wof b3 is2 y z x
Beam divergence ()
is1 wco
b)
Fig. 10. Attenuation calculated by the ray-tracing method for the old and the new fluxmeter. CONCLUSIONS
We have described a fluxmeter for solar concentrators, whose main characteristics are: the attenuation of radiation is allowed in a simple and precise way; flux density from few suns to thousands of suns can be measured (four decades of dynamic range); the response is fairly independent of the angular divergence of input beam; it is equipped with an external spectrometer for spectral measurements; measurements are carried out under stationary irradiation regime; the photodetector is thermoregulated for having a constant sensitivity; easy to operate and portable.
b2 wpd
Fig. 8. Vertical (a) and horizontal (b) sections of the gauge sensor.
Now on we analyze by raytracing the response of fluxmeter at beams with different divergence at constant power in input, in presence or not of the three baffles. Fig. 9 shows, as expected, an average attenuation higher with baffles, due to the transfer of light from (is1) to (is2) limited by (b2) and (b3), which act as reflectors. Fig. 9 shows also two fluctuations of f A : one increase at ~50 in presence of baffles, due to the reflecting effect of (b1) respect to input radiation from (win); the other at ~40 in absence of baffles, due to the direct input of light from (win) into (is2). Maximum fluctuations of f A are around 5%. These fluctuations have been strongly reduced by modifying the internal configuration of the two cavities. In Fig. 10 we report the final results, showing that the fluctuations, lower than ~0.5%, have been reduced of one order of magnitude.
65
REFERENCES
[1] A. Luque, G. Sala, J.C. Arboiro, T. Bruton, D. Cunningham and N. Mason. Some results of the EUCLIDES photovoltaic concentrator prototype. Progress in Photovoltaics: Research and Applications 1997. 5, pp. 195-212. [2] I. Antn, D. Pachn and G. Sala, Characterization of Optical Collectors for Concentration Photovoltaic Applications, Progress in Photovoltaics: Research and Applications 2003. 11, pp. 387-405. [3] A. Parretta, C. Privato, G. Nenna, A. Antonini, M. Stefancich, Monitoring of concentrated radiation beam for photovoltaic and thermal solar energy conversion applications, Applied Optics 2006. 45 pp. 7885-7897. [4] A. Ferriere, and B. Rivoire, Measurement of Concentrated Solar Radiation: The Asterix Calorimeter, in Proc. 10th SolarPACES Int. Symposium on Solar Thermal Concentrating Technologies, Sydney, 8-10 March 2000, H. Kreetz, Ed. (ANU Australia), pp. 233-240.
Attenuation factor, fA
60 55 50 45 40 10 20
baffles no baffles
30
40
50
60
70
80
90
Beam divergence ()
Fig. 9. Attenuation factor as function of input beam divergence.
134
AUTONOMOUS POLYGENERATION SOLAR CONCENTRATOR

Jon Bhmer Kyoto Energy Ltd. Astridsvei 3 0276 Oslo Norway jon@kyoto-energy.com
Abstract Kyoto Energy has designed a novel solar concentrator that can be used for a number of different tasks by utilizing both the heat, light and electricity potential in solar energy. It is designed for use in developing countries, where autonomous energy systems are much needed and must use solar energy, which is the most abundant source.
cost and easiest manufacturing could be achieved. The optical principle chosen was Cassegrain, which even with two mirrors allowed a system efficiency of 94% and also allowed the receivers to be mounted at the balance point of the unit as opposed to at the primary focal point. The mirrors were tested an yielded a 10cm beam, allowing for 280x concentration. The use of separate sections allowed the system to be transportable and collapsible, as well as being easy to assemble by users.
Central Conclusions Tracking System An system has been devised which is robust, low cost, portable and very flexible. The novel use of existing mass manufacturing processes allows it to be produced in most countries. A single system as for the first time been devised to allow cooking, electricity, hot water/air, cooling and daylighting. A small electrical brushless DC motor was employed, using 0,1W on average. Heat sensors on either side of the concentrated beam as well as a fixed solar speed basic algorithm and end sensors provide the necessary tracking information. Separate small solar panel provides electricity for motor. The electronics and motors have an estimated 20 year MTBF.
Low cost, robustness To lower costs and ease distribution and transportation, we aimed to use as little materials as possible. The choice of aluminum and plastic ensures light weight, low costs and it is manufacturable in many existing factories around the world. To find the right size of unit the necessary energy for cooking for a family was estimated at around 2kW thermal. This equates to mirrors about 2,25m2 in size, but this can be sized up or down at will. Our findings also indicate that there are few reasons to make larger units rather than simply making more units, unless the application is industrial high temperature ones.
Receivers Based on the above generic solar concentrator, a number of receivers have been designed. Cooking a reflective pipe allows cooking next to the unit or inside a house Hot water and hot air the same receiver unit can be used for both purposes Electricity/hot water co-generation water is heated while cooling the PV cells Daylight use of a fiber optic cladded/non-cladded system allow daylight inside buildings Cooling a special ice maker with no moving parts is designed Small-power Stirling provides mechanical energy and/or electricity
Optical subsystem The most critical part is the optical subsystem. We needed a reflective material with the highest performance and long life, and found a polymer film with 99% reflectivity in the visual and 20% in the IR spectrum. By adding a UV layer it is estimated to last for about 5 years. By using a lamiation technique in plastic injection molding, the lowest
Applications Electricity about 600w per unit Cooking about 2kW per unit
135
Hot water either separate or as a byproduct of electricity generation Crop drying using forced hot air with a temperature regulator and a separate drying box Cooling using a special Pasteurization heating water to 68C before ejecting from receiver using a special valve Desalination providing electricity for pumping through membrane Water Pumping in combination with a special lowcost deep well water pump of our design Mechanical work (stirling or PV/motor) for workshops Ethanol production solar assisted 400L daily production using rotten fruit BioDiesel production solar assisted
136
PRACTICAL COGENERATION WITH CONCENTRATING PV

A. Kribus and G. Mittelman School of Mechanical Engineering, Tel Aviv university, Tel Aviv 69978, Israel
ABSTRACT
In photovoltaic systems most of the collected radiation energy is converted to heat and lost to the environment. Cogeneration, capturing and using this wasted thermal energy, is a way to increase the overall efficiency. We consider the practical potential performance, based on properties of currently available technology. A semiempirical model is developed to describe the performance of a triple junction cell as a function of concentration and temperature. Several concentration levels representing common concentrator technologies are investigated. Several paths for the use of the waste heat are considered. An overall equivalent efficiency is used to quantify the combined value of the two energy streams provided by cogeneration. The results show that for some practical scenarios, cogeneration can provide a significant advantage relative to the generation of electricity alone.
reduce the electric conversion efficiency of the PV cells. On the other hand, increasing the incident flux on the cells increases the efficiency. Therefore, operation of PV cells can be considered at temperatures that would be unacceptable without concentration. A model of PV cell performance as a function of temperature and concentration is necessary to analyze these effects. Published theoretical values of triple junction cell efficiency under concentration [7, 8] do not match experimentally reported data [9-11]. We have developed a semi-empirical model that can be reasonably fitted to the experimental data, as a basis for evaluation of a PV cogeneration.
INTRODUCTION
The conversion efficiency from sunlight to electricity for photovoltaic systems is usually in the range of 1030%. Concentrator PV cells with efficiency around 40% have been demonstrated, but system efficiency with these cells, including concentrator and other losses, will still be around 30%. Most of the solar energy is therefore rejected back to the environment. Most of this waste energy is available as heat that can be collected and converted to a useful energy product, increasing the overall efficiency. This is cogeneration: providing more than one energy product in parallel from the same initial resource. The easiest path for cogeneration is to provide the waste heat directly to an end user that requires, for example, hot water or space heating, as shown in Figure 1(a). This is done by Photovoltaic/Thermal (PV/T) flat plate collectors [1] that contain a heat exchanger behind the PV cells for collection of the thermal energy. Flat plate PV/T systems, however, are capable of providing only lowtemperature heat up to about TU=60C. Concentrating Photovoltaic Thermal (CPVT) systems may provide higher temperatures and therefore a wider range of applications: not only hot water [2], but also steam and industrial process heat. Theoretically it is also possible to convert the waste heat from PV cells into additional electricity, as shown in Figure 1(b), using a hypothetical Carnot engine [3] or a more realistic engine [4]. The waste heat may also be further converted into other end products, such as cooling, Figure 1(c) [5] and desalinated water [6]. In order to make the thermal energy more valuable, it is necessary to increase its temperature. However, this will Figure 1: Solar cogeneration: (a) heat to user, (b) Heat Engine (HE) to generate additional electricity, (c) Absorption Heat Pump (AHP) to generate cooling Cogeneration produces two types of output energy that are not equivalent, and the definition of overall efficiency is not always unique. We have to account for the different values of heat and electricity, and for the different value of heat at different temperatures. The selection of a comparison metric depends on the application. In this work, we analyze the potential for increasing the overall efficiency of photovoltaic systems by cogeneration. We model realistic systems rather than finding the theoretical upper limit of performance, by using the properties of an actual triple-junction PV cell, and a model of typical losses (optical, thermal) from a solar collector system. Two cases are investigated: PV conversion with a bottoming heat engine (HE), and PV conversion with a bottoming absorption heat pump (AHP).
PV receiver model
The PV cell performance is based on reported measurements of an InP/InGaAs/Ge triple junction cell [7, 10]. The cell equivalent circuit includes three single-diode subcell elements connected in series, as shown in Figure 2. The current-voltage relationship for each subcell is:
q(Vi +JL ARs ,i )
JL = Jsc,i Jo,i (e
ni kBT
1)
Vi + JL ARs,i ARsh,i
(1)
137
where JL is the current at the cell terminals; Jsc,i is the short circuit current; Jo,i is the reverse saturation current; q is the electric charge; Vi is the voltage; A is the cell area; ni is the diode ideality factor; kB is the Boltzmann's constant; T is the absolute temperature; Rsi, Rshi are the serial and shunt resistances, respectively.
model parameters was done using the data of [8-11]. The 2 resulting values are: Rs=0.102 /cm , =0.00046 eV/K, =204 K, and the subcell parameters shown in Table 2. Table 1: Known model parameters Subcell Eg (eV) JSC (mA/cm2) 1 1.82 13.78 2 1.40 15.74 3 0.65 20.60
Table 2: Calibrated model parameters Subcell n (mA/cm2K4) 1 1.9 9.3109 2 2 1.5 6106 2 3 1.4 2.1105 2
The dependence of cell efficiency on concentration and temperature is usually expressed as:
C = o (C ) + (C )(T 25)
(6)
Figure 2: A triple junction cell: structure [10] (left), equivalent electrical circuit (right) The overall voltage of the stack is then, neglecting the effect of the shunt resistances [9], the sum of subcell voltages: V= J J kBT 3 n ln( sc,i L +1) JL ARs Jo,i q i=1 i (2)
where o is the cell efficiency at 25C and is the temperature coefficient of the efficiency. The model results as a function of C, compared to available experimental data, are shown in Figure 3. The model shows a good fit to the experimental data, in contrast to the theoretical prediction given in [7] that shows a significant overprediction. Maximum efficiency is near C=300, consistent also with data from Spectrolab for similar cells. Based on these results, the baseline efficiency at 298 K may be approximated by the following simple correlation:
o =
0.295 + 0.0157ln(C)
8 2 0.364 5 10 C + 0.000048C 100 < C 1000
1 C 100
(7)
where Rs is the sum of all series resistances. The cell efficiency is defined by dividing its output power by the incident power on the cell:
C = JLV CG
(3)
where G is the incident flux at the concentrator aperture and C is the flux concentration ratio. The current and voltage Jm, Vm at the Maximum Power Point (MPP) are obtained by: d(JLV)/dJL=0. For non equal diode ideality factors, this equation is solved numerically for JL. Some of the model parameters depend on the cell temperature and the incident flux. The short circuit currents JSC,i are proportional to the flux. The reverse saturation currents are temperature dependent [12]:
Jo,i = i T
(3+ i /2)
E g ,i ni kBT
(4)
Figure 3: Cell efficiency vs. concentration, model fit to published empirical data Cell efficiency vs. temperature is shown in Figure 4, compared to experimental data. The model predics the linear trends and reasonably accuracte efficiencies. However, the predicted slope (temperature coefficient) at high concentration is inaccurate: the model predicts a faster decline with temperature (=0.057%/K for C=200) than the measured value (=0.0363%/K). In order to resolve the dependence of the temperature coefficienct on concentration, a correlation was constructed from the model results, and the correlation coefficients were then re-calibrated to fit the few available empirical results.
where , are constants. should be between 0 and 2. The bandgap Eg also depends on temperature [13]:
E g,i = E g,i (0) T 2 T +
(5)
where and are constants that may be set to the same value for all subcells. Some of the needed parameters are available for all subcells [9] for T=298 K and C=1, as shown in Table 1. The other parameters need to be found by fitting the model equations with empirical data. This calibration of
138
The overall system efficiency, including electricity output from both the PV cells and the heat engine, is:
= opt PV + opt ( PV Lrec ) K ( A TPV ) 1 1T
(11)
Figure 4: Cell efficiency vs. temperature, model fit to published empirical data
( )= a1 + a2 ea C C
3
(8)
where opt is the optical loss in the concentration system, and Lrec is the receiver thermal loss [14]. The overall system efficiency as a function of temperature and concentration ratio is shown in Figure 6. For C=1 clearly cogeneration does not increase the efficiency. For C=50, an optimal temperature of about 200C is seen, with overall efficiency of 33.2%: 22% more electricity than the CPV alone at room temperature. For C=500, the optimal temperature is about 400C, which is clearly impractical for a PV receiver. At 250C the system efficiency is 38.0%, 33% more electricity than the same CPV system operating at room temperature.
0.6
The fit is shown in Figure 5. Based on this fit, it is finally possible to derive the cell efficiency at any concentration and temperature, using equations (6), (7) and (8). For example, the following values represent typical situations of no concentration, linear concentration, and point-focus concentration:
0.295 0.000730(T 25) C = 1 C = 0.357 0.000376(T 25) C = 50 0.376 0.000363(T 25) C = 500
0.4
C=500
C=50
0.2
C=1
(9)
0.0 0 100
Temperature TH (C)
200
300
Figure 6: Efficiency of CPVT system with a bottoming HE

CPVT with AHP
Figure 5: Fit of model and empirical data to correlation (8)
CPV system performance

CPVT with Heat Engine We consider a Concentrating PV/Thermal (CPVT) system with a heat engine accepting the waste heat from the PV cells. The heat engine operates at K=0.6 of Carnot efficiency, typical of realistic engines. This is similar to the configuration studied in [3, 4]. The PV module efficiency is lower than the individual cell efficiency due to gaps among the cells and current mismatch loss, which we estimate as 10% or mod=0.9:
The heat collected from a CPVT collector can be fed into an Absorption Heat Pump (AHP) to produce air conditioning or refrigeration. This is an attractive solar application since the need for cooling is well correlated with the availability of solar radiation. AHP systems are commercially available, with typical performance of COPAHP=0.7 and 1.2 for single-effect (SEAHP) and double-effect (DEAHP) systems, respectively. In contrast to heat engines, the COP of an AHP does not change much with temperature and therefore we assume that it is constant within the relevant ranges: 0.7 between 60C (threshold for operation of a single-effect system) and 150C, and 1.2 above 150C (threshold for operation of a double-effect system). The overall electric efficiency in this case includes the electricity produced by the PV cells, and the electricity that would have been supplied to a conventional compression cooler that would produce the same amount of cooling power. Typical performance of a reference conventional cooler is COPref=5. The overall efficiency is then:
= opt PV + opt ( PV Lrec ) 1
COPAHP COPref
(12)
PV = C mod
(10)
The results for the CPVT system with bottoming AHP are shown in Figure 7. The best efficiency for C=500 with a SEAHP is 35.7% at 60C, corresponding to 25% more electricity than the CPV alone at room temperature, and
139
14% more electricity than the CPV+HE option. For DEAHP system, the best efficiency is 39.4% at 150C, 37% more electricity than the CPV alone and 11% more than the CPV+HE option at the same temperature. The results for C=50 are similar, with best cogeneration efficieny of 34.1% and 36.3% for the SEAHP and DEAHP, respectively, compared to 27.2% efficiency for electricity alone at the same concentration.
REFERENCES
1. Van Helden, W.G.J., R.J.C. van Zolingen, and H.A. Zondag, PV Thermal systems: PV panels supplying renewable electricity and heat. Prog. PV Res. Appl. 12, 415426, 2004. Coventry, J.S., Performance of a concentrating photovoltaic/thermal solar collector. Solar Energy 78(2), 211-222, 2005. Luque, A. and A. Mart, Limiting efficiency of coupled thermal and photovoltaic converters. Solar En. Mat. Solar Cells 58, 147165, 1999. Vorobiev, Y.V., J. Gonzalez-Hernandez, and A. Kribus, Analysis of potential conversion efficiency of a solar hybrid system with high temperature stage. J. Solar Energy Engrg. 128, 258-260, 2006. Mittelman, G., A. Kribus, and A. Dayan, Solar Cooling with Concentrating Photovoltaic/Thermal (CPVT) Systems Energy Convers. Mgmt. Submitted, 2006. Mittelman, G., A. Kribus, and A. Dayan. Desalination with Concentrating Photovoltaic Systems. in Solar World Congress. 2005. Orlando: ISES. Yamaguchi, M., T. Takamoto, and K. Araki, Super high-efficiency multi-junction and concentrator solar cells. Solar Energy Materials & Solar Cells 90, 30683077, 2006. Yamaguchi, M., T. Takamoto, and K. Araki. High efficiency concentrator multi-junction solar cells and modules. in ISES Solar World Congress. 2005. Orlando. Nishioka, K., et al., Evaluation of InGaP/InGaAs/Ge Triple-junction Solar Cell and Optimization of Solar Cell's Structure Focusing on Series Resistance for High-efficiency Concentrator Photovoltaic Systems. Solar En. Mat. Solar Cells 90, 1308-1321, 2006. Nishioka, K., et al., Annual output estimation of concentrator photovoltaic systems using highefficiency InGaP/InGaAs/Ge triple-junction solar cells based on experimental solar cells characteristics and field-test meteorological data. Solar En. Mat. Solar Cells 90, 5767, 2006. Nishioka, K., et al., Evaluation of temperature characteristics of high-efficiency InGaP/InGaAs/Ge triple-junction solar cells under concentration. Solar En. Mat. Solar Cells 85, 429-436, 2005. Meillaud, F., Efficiency limits for single-junction and tandem solar cells. Solar Energy Materials & Solar Cells 90, 2952-2959, 2006. Luque, A. and S. Hegedus, Handbook of Photovoltaics Science and Engineering. 2003, New York: Wiley & Sons, Inc. Kribus, A. and G. Mittelman, Potential of polygeneration with solar thermal and photovoltaic systems. J. Solar Energy Engrg., In Press, 2007.
2.
3.
PV+AHP
0.4
4.
PV+Engine
0.2
PV
PV
0.0 0 50 100 150 200 250 300
5.
6.
Temperature (C)
PV+AHP
0.4
7.
PV+Engine
0.2
8.
PV
PV
0.0 0 50 100 150 200 250 300
9.
Temperature (C)
Figure 7: System efficiency of a CPVT system with a bottoming AHP: (a) C=50, (b) C=500
10.
CONCLUSIONS
The analysis of cogeneration with CPV systems shows that the overall efficiency of a cogeneration system can be significantly higher (around 40%) than the CPV system generating only electricity (less than 30%). The increase in temperature, needed for effective cogeneration, reduces somewhat the conversion efficiency of the PV cells. However, the gain due to increased quality of the thermal energy compensates for this reduction, and leads to a significant gain in equivalent electricity displacement. Obviously, the increased temperature requires concentration of the incident sunlight. Comparison of electricity generation with the waste heat vs. thermal use (absorption cooling) shows that the thermal use can lead to higher overall efficiency. The long-term operation of PV cells at elevated temperature may raise some concern. Such operation has been reported at 240C but it is not clear what are the long-term effects on cell performance. This is an open question that should be investigated. 11.
12.
13.
14.
140

Session 3A 1

Transféré par

Informations du document

Description originale:

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Session 3A 1

Transféré par

Droits d'auteur :

Formats disponibles

SESSION 3A VISUAL PRESENTATIONS

COMPARISON BETWEEN DIFFERENT SOLAR CONCENTRATORS AS REGARDS TO THE ELECTRIC GENERATION

Solar technologies description

Fig. 1 Functional diagram of a STC solution.

High PV Concentrator (HCPV)

field = mirror * htf * power cycle

0,626 0,089 0,224 2,574 0,045 268 0,147

24,3 3,5 8,7 100 1,7

0,501 0,072 0,179 2,375 0,036 335 0,131

21,1 3,0 7,5 100 1,5

Low concentrator (LCPV)

0,12 0,10 0,08 0,06 0,04 0,02 0,00

STC curve cost

0,12 0,1 0,08 0,06 0,04

0,559 0,102 0,200 1,641

34,1 6,2 12,2 100,0 3,1

0,556 0,101 0,200 1,445 0,051 237 0,113

38,4 7,0 13,8 100,0 3,5

0,573 0,104 0,200 1,498 0,052 230

38,25 6,95 13,35 100,00 3,48

0,02 0 0 50 100 150 200 250 300 350 P ow e r [M W ]

Fig 3 Cost of the system vs. installed power

0,08 0,06 0,04 0,02 0,00

Concentrar factor [GaAs]

OPTICS and ENCAPSULATION CONCENTRATOR SOLAR CELL

cell size [mm] concentrator factor [20 100 200]

X=108 X=167 X=242 X=310 X=383 X=445 X=518 X=588

cell efficiency (from est. optical losses)

-0,08 -0,1 -0,12 -0,14 -0,16 -0,18 -0,2

concentration ratio (X)

-0,01 X=40 -0,015

-0,02 X=73 -0,025

Racp 1<109 109 < Racp 1

and for the rear surface (cp-2) of the heat sink :

1/ Nu cp 2 =0.27Racp4 2 105 < Racp 2 <1011

Concentration ratio 1000 2000

HIGH-FLUX CHARACTERIZATION OF ULTRA-SMALL TRIPLE-JUNCTION CONCENTRATOR SOLAR CELLS

performance for broad ranges of anticipated operating conditions.

LIP: 0.6 mm fiber

b G I 2 = I 1 + I sc1 I 2 1 + [T2 T1 ] 10 1 1000 G I0 0 1

= Pm/Pin = FF Isc Voc/Pin

Kaleidoscope LIP: 1.0 mm fiber LIP: 0.6 mm fiber Flash simulator

RESULTS AND DISCUSSION

Voc (nkT/q) ln(Isc/Io) (nkT/q) ln(Pin) + const , (3)

FLUXMETER FOR PARABOLIC TROUGH SOLAR CONCENTRATORS

BASIC SCHEME OF THE FLUXMETER

Ei(c ) , measured orthogonally to the average direction ()

Ei( r ) [ z ( )] = cos Ei( c ) [ z ( r2 r1 ) tg ] ( r2 / r1 ) (2)

... = tot r1 z i Ei( r ) / N

tot [ z , z ] = tot r1 z E ( r ) ( z ) = ...

radians. The same considerations are valid if

Fig. 5. Section of a sensor head window.

below the maximum temperature recommended by the manufacturer (100 C).

Signal / Rad_eff (arb. u.)

10:00 azi = -32.0 deg CFV2 ideal AC01

0,5 10 0 -100 -50 0 (deg) 50 100 0,0

Inverters response time with concentration PV systems

eta DNI GNI power

600 DNI - GNI - Power

Fig. 2: Data logging system scheme

eta DNI GNI power 1000

16,0% 14,0% 12,0%

600 10,0% 8,0% 400