THE DESIGN AND TESTING OF A LOW-MATERIAL-COST PARABOLIC-TROUGH PV CONCENTRATOR

Transcription

1 THE DESIGN AND TESTING OF A LOW-MATERIAL-COST PARABOLIC-TROUGH PV CONCENTRATOR Clive K.Weatherby Dept. of Cybernetics, University of Reading, Whiteknights, PO Box 225, Reading, Berkshire, RG6 6AY, UK. Tel , Fax , C.K.Weatherby@reading.ac.uk ABSTRACT: This paper describes the development of a polar-axis tracked small-aperture parabolic mirror PV concentrator designed to minimise both material usage and production costs. The research follows an earlier JOULE II funded study (EUCLIDES), as part of which a number of collectors in current production in the US were examined, and their performance and cost estimated. The collector described in this paper has two parabolic trough mirrors in a single rectangular box. The trough width is 20cms so that the casing material is an adequate heat sink to cool the cells. It has a geometric concentration ratio of 19.2, uses a reflective self-adhesive coating on a thin stainless steel parabolic substrate, and employs BP s Laser buried grid (LBG) monocrystalline cells optimised for up to 40 suns cut to 10mm x 48mm. The collector has a plain glass cover for easy cleaning and protection of the optical surface. The collector body was constructed from 1mm duralloy sheet and has an aperture of 1m x m. In volume production, it is suggested that the collector body can be pressed into an accurate profile and be used for the parabolic mirror substrate. Due to its low weight and the box type construction, the substrate would not distort easily. Construction experience and performance results show that this collector is practical and that material savings are achievable. Testing of the collector is continuing. Keywords: Concentrator - 1 : PV Module - 2 : Pricing INTRODUCTION There are many competing concentrator PV systems and it remains a difficult problem to choose the most cost-effective variant for mass production. An earlier JOULE II study revealed that current systems in production in the US were designed with relatively large apertures thus requiring more rigidity and hence more material to be self supporting. It is thought that this was to reduce the labour costs associated with small production volumes. It is believed by the author that future reductions in the cost of concentrator systems will come as a result of large scale investment in tooling and by careful product design for large scale manufacture. It is further believed that if large scale production is realised, the material costs could be as much as 2/3 the final system cost; therefore it is the amount of material which should be minimised at the design stage. In addition, it was found during earlier research [1] that reducing the aperture of a concentrating collector has the benefit of allowing passive cooling from the collector body without additional heat sinking. For these reasons, a small-aperture parabolicmirror concentrator, with a minimum of material was proposed. Figure 1: A general view of the small mirror collector. The collector has been incorporated into an ongoing JOULE programme where the aim is to study a wide variety of low to moderate concentration collectors in search of cost-effective solutions [2]. The collector described here is designed to have a concentration ratio suitable for use with BP s LBG cells optimised for up to 40 suns [3]. It is believed that for large volume production, the cost of these cells will be little more than that of one-sun cells and can therefore offer a cost benefit at moderate concentration ratios. It was clear, from the data of the US visit, that high optical element efficiency was considered to be an important factor in reducing energy cost, and research [2] has shown that it is reasonable to increase primary optical element efficiency despite a cost penalty. For this reason 3M s ECP305+ silver coated PMMA film was chosen for the primary optical surface since, although expensive, it has a high solar reflectance (up to 94%) over the bandwidth of

2 interest. Polar axis is the tracking strategy initially selected. 2. COLLECTOR DESIGN 2.1 General guidelines used in the design Use optical elements as part of the housing providing rigidity and cooling in the case of a mirror, or protection and rigidity in the case of a lens. Where possible use the housing to cool the cell strings; this requires the aperture to be small if the housing material is not too thick. Use the basic structural elements already present in the housing to provide any rotating axis pivots. Use a concentration ratio which minimises the cost per watt output of the collector. This usually means choosing cells with optimum efficiency at the average concentrated incident light energy level. Avoid cell strings which are complex to manufacture; this is a problem for small cells but with automation this problem may be lessened. Using these guidelines, the prototype small mirror collector pictured in figure 2 was designed. The design was predicated on the use of the BP LBG cells which were optimised for 40 suns. A guard band of half the cell width was allowed to extend the acceptance angle and to offset any mirror slope errors. Cells were chosen to be 1cm wide based on the need for a maximum of 40 suns at any point on the cell string. The geometric concentration ratio was thus set at about Cooling Earlier research into the optimum length of thin cooling fins [1] showed that if the temperature above ambient of a concentrator s receiver was to be kept to that of a typical planar array with only a 1mm thick housing as the heat sink, then the aperture should be restricted to about 20cms. It was evident that with a glass cover, some heat from the cells would be retained within the collector. After further numerical analysis it was decided to add protruding fins which would be exposed on two sides to aid cooling. 2.4 Mounting and tracking It has been demonstrated by the SEA Corporation [5] that if modules are sufficiently lightweight in construction, it is possible to mount several of them on a simple frame tilted at the latitude angle and to rotate them from East to West with a single inexpensive motor. For the collector described here, control is effected by an inexpensive drive circuit with a single sensory head employing a mixture of analogue and digital techniques. 3. RAY-TRACING Figure 2: The small mirror collector cross section. 2.1 The mirrors It has been argued that parabolic mirrors are not well suited to photovoltaic applications because they are prone to surface slope errors and distortion under wind loads. Furthermore, since small slope errors of the mirror lead to twice the deflection of the incident radiation, then the reluctance to employ this technology can be understood. However, it has been demonstrated by UPM [4] that with sheet aluminium mirrors and the right manufacturing techniques these problems can be largely overcome. By reducing the collector aperture the amount of material required to maintain the rigidity of the mirror reduces to the extent that material thickness is dictated mainly by its ability to conduct the heat from the cells. The glass cover which is glued to the body provides additional rigidity. 2.2 Cells 3.1 Method Ray tracing was carried out on the basic conceptual design using software from the Opticad corporation which allowed 3D analysis of optical concentrator designs for imaging and non-imaging applications. The software uses non-sequential ray tracing techniques. Multiple refractions and reflections can be viewed and energy distribution across an image plane can be saved for further analysis. In general the analysis was carried out using macros, with the collector and solar source created separately. Movements of the collector were simulated by changing the direction of the solar model. The latter consisted of nine weighted collimated sources, each at a different angle to the collector, representing the variation in intensity across the solar disc. A small amount of circumsolar radiation was incorporated into the model. Movements were effected by dedicated macros. The shape and positioning of the collectors components were changed to allow various permutations to be assessed. The mirror surface parameters were not known so details of a Flabeg glass mirror made for the Luz installation in Arizona were used as a guide. Tests of one of these mirrors showed that the specular reflections occupied a cone of roughly degrees. The mirrors of the

3 model were given a Gaussian scattering function with the standard deviation set to degrees. This was therefore much worse than the Flabeg case. The 3M ECP 305+ reflective material has a surface coating of PMMA which acts as a protective layer. The refraction and secondary reflections created by this layer were disregarded as the effect was not considered significant. The reflection coefficient was set to 94% which is consistent with 3M s published data. The cell string was modelled with a film plane set to absorb all the incoming radiation. Superimposed on this was a glass cover with a refractive index of 1.5 to simulate the encapsulent. The cell string was aligned at 45 degrees from the normal to the collector aperture (see figure 2). 3.2 Results of the ray-tracing analysis The ray tracing revealed an asymmetrical distribution of energy on the cell string when the collector was normal to the sun (figure 3). This distribution is not ideal and changing the mirror profile could make this distribution more uniform. Step changes in the angle of incoming radiation allowed the acceptance angle to be studied. The results revealed that 85% of the energy could be captured if the collector remained within ±5 degrees of the optimum position and 90% captured if the collector remained within ±5 deg. (figure 4). Note: The results do not take account of the Fresnel losses at the module s glass cover to air boundaries Energy percentage Energy distribution accross a Si solar cell at the focus of the small mirror collector with a degree rotation in the solar model Y position on cell (mm) Figure 3: Ray-trace energy distribution across a receiver cell Energy capture (percent) Energy capture versus angular misalignment for a silicon solar cell at the focus of the small mirror collector. Modelled in opticad 17/5/ E/W angle of rotation of the solar model (degrees) Figure 4: Energy capture versus solar incidence angle (from the ray-trace analysis). 4. THE PROTOTYPE The prototype was constructed by first fabricating a box section. The mirror was formed separately by hand using thin gauge, work-hardened stainless steel sheet, supported by ribs which had been cut from plate duralloy by a CNC machine. Figure 5: Cross section of the proto-type. A self adhesive reflecting material was applied to the stainless steel prior to mounting on the ribs. The exact spectral performance of the film used is not as yet known but tests are ongoing. Although the construction method was not the same as that envisaged in final production, it was thought that the prototype would exhibit a similar enough behaviour for useful analysis. By reference to a reflected image, the mirror slope errors were thought to be large. The cell string was made from 16 BP LBG cells which had been laser cut from 48mm x 96mm cells into 48mm x 14mm pieces (with an active area of 48mm x 10mm) as shown in figure 6. Figure 6: Details of how the cells were cut.

4 A batch of these cells were separately assessed for their I/V characteristics. Calibration of the test equipment was made using a master cell which had been calibrated at NREL under the ASTM E 892 spectrum. Subsequent checks against the original BP data for the cells in their complete, uncut form showed good agreement. Cells were then selected by matching their I/V curves. The cells were mounted on a machined aluminium bar insulated by Chomerics Thermattach T404 double-sided self adhesive tape. Each cell was tabbed in a way which allowed subsequent analysis of the individual cells as can be seen in figure 7. figure. First tests of the reflective material reveal a total specular reflectance for the wavelengths of interest to be about 85% which alone would reduce the concentration ratio from 19.2 to It is believed that surface slope errors account for much of the balance. The collector was connected to a PC-based data logger and an electronic load driven by a D to A card. This enabled the acceptance angle, tracking accuracy and I/V characteristics to be measured Short circuit current versus time/angle for the small mirror collector. Global and diffuse insolation were reasonably constant. 17/10/ Time (secs) Gridlines at intervals of 1 minute =.25 degrees Current (Amps) Figure 7: The cell string for the prototype. An additional cell (previously calibrated ) was placed at the centre of the cell string and connected separately from the rest. It therefore did not form part of the main string. This was to enable a separate record of the performance without any distortion of the results from cell matching. A similar calibrated cell will be shaded so as to intercept only beam radiation. It will be mounted close to the cell string and thus will be at a similar temperature. Data logging of the short circuit current from these cells will allow instantaneous concentration ratio readings. 5. RESULTS Initial tests of Isc were carried out by hand on each cell in the string under concentration (figure 8). Isc-Cells in the string (A Short circuit current for each cell in the string of the small mirror collector in normal sunlight whilst tracking on a polar axis. Collector at the latitude angle. 24/9/ am BST Cell Isc normalised to beam radiation Cell number. (No9 is the calibrated centre cell and is not in the string). Figure 8: The cell string under concentration. These tests revealed some variation in the short circuit current along the string indicating some change in the quality of the mirror over its length. Despite the poor quality of the mirror it was possible to assess the performance of the collector using the data from the central cells which gave more consistent results. The concentration ratio calculated from the data at the calibrated centre cell was 13.9 as opposed to the 19.2 ideal Figure 9: Short circuit current versus hour angle. It was also possible to initiate peak power point tracking using I/V data taken at pre-set intervals e.g. 10 seconds. A full I/V curve with several hundred readings could be taken in under 1 second. A typical I/V curve is shown in figure 10. Current (A) IV characteristics of the cell string of the small mirror collector under concentration. 17/10/96 at 2.15pm Each data point is an average of 100 readings. IV curve Power curve Voltage (V) Figure 10: I/V curve for the small mirror collector. The I/V curve was surprisingly good considering the earlier mirror problems and it shows well matched cells. The fill factor was over 80%. It is fair to assume that the current was set by the worst case cell and the earlier graph (figure 8) would appear to suggest that up to 25% more power could be gained if the mirror were more accurate. The efficiency was calculated by logging the power output of the collector together with global and diffuse radiation. It was calculated at 9.2% (with cover). Winds were very light, ambient temperature was about 18 C and the cells were at about 46 C. The average working temperature is expected to be about 35 C Power (W)

5 Table 1: Cost per peak watt for a small selection of solar concentrators manufactured in large volumes which were considered under the current research programme [2]. Flat Panel at Latitude Point - Focus Fresnel cpc 330x 2-axis Curved Cylindrical Fresnel Polar The Small Mirror Collector Polar $4.31 $2.68 $2.45 $1.62 [3] N.B.Mason, T.M.Bruton, K.C.Heasman, Proc. 13th European PV Solar Energy Conf., H.S.Stephens, Bedford, (1995), [4] J.C.Arboiro, G.Sala, I.Molina, Proc. 13th European PV Solar Energy Conf.,H.S.Stephens, Bedford, (1995), [5] N.Kaminar, J.McEntee, P. Stark, D.Curchod, CH2953-8/91/ IEEE 1991 pp Although the results from the spreadsheet analysis shown in the above table are very general, a clear indication is given of the collectors which merit further study. The collector referred to here may therefore warrant further research but other collectors also show promise. 6. FUTURE WORK New mirrors are planned which will give better energy distribution on the cells and a reduction in the slope errors when compared to the current hand made mirror. The ECP305+ material from the 3M corporation has been discontinued and SS95P (a product with similar reflective properties) has been selected as a replacement. This product is not intended for direct exposure to the weather but since the collector is covered it is expected that this cheaper product will be adequate. The annual average output for varying sites will be modelled and the system parameters will be optimised for further cost reductions. 7. CONCLUSIONS It has been established that this is a promising collector. The collector s acceptance angle is large enough to allow low-cost tracking methods and a good degree of tolerance to mirror profile errors. Cost analysis done as part of related research [2] indicates that it has low material costs and low costs per kw hour. The prototype obtained an efficiency of 9.2% but used a mirror of poorer quality than desirable. However it is anticipated that the efficiency with a more uniform mirror will be circa 12% and with a better reflective surface, over 13%. Further optimisations are proposed and quality will be improved to meet the desired efficiency levels. REFERENCES [1] G.R.Whitfield, R.W.Bentley, C.K.Weatherby, Proc. 13th European PV Solar Energy Conf., H.S.Stephens, Bedford, (1995), [2] G.R.Whitfield, J.C.Miñano, H.D.Mohring et al, 14th European PV Solar Energy Conf. (1997).