Sie sind auf Seite 1von 11

Solar Energy 74 (2003) 417427

The effective size of the solar cone for solar concentrating systems
D. Buie*, C.J. Dey, S. Bosi
Solar Energy Group, School of Physics A28, University of Sydney, Sydney, NSW 2006, Australia Received 15 December 2002; received in revised form 10 April 2003; accepted 21 April 2003

Abstract In this paper we dene a virtual solar cone, whose principle axis is aligned with the solar vector, having a radial angular displacement containing a pre-dened proportion of the terrestrial solar radiation. By simulating various sunshape proles, the angular extent of the energy distribution is established to give the effective size of the solar cone for a range of atmospheric conditions. Then, by simulating the reection of these solar distributions off a set of non-ideal mirrored surfaces, accounting for non-specular reection and mirror shape errors, the combined effect of sunshape and mirror properties on the solar image is obtained. Clear trends are presented that show the dependence of the effective size of the solar image on the accuracy of a mirrored surface for different sunshapes. We then identify the effective size of the solar image at the absorber plane that must be accommodated in the design and optimisation of solar concentrating systems. 2003 Elsevier Ltd. All rights reserved. Keywords: Insolation; Concentrating systems; Sunshape

1. Introduction The broadening of the solar image when sunlight travels from the near vacuum of space to the Earths surface is due to interactions of the sunlight with atmospheric particulates. Where the radius of the particles is large in comparison to the wavelength of the sunlight, small angle forward scattering occurs, forming the solar aureole (Mie, 1908). The amount of energy in this circumsolar region is important for two reasons (Noring et al., 1991; Buie et al., 2003). Firstly, the radial solar energy prole (sunshape) plays a non-negligible role in determining the overall ux distribution at the focal region of imaging concentrators. Secondly, depending on the acceptance angle of a solar concentrating system, overestimation of the power output may occur if all the power is assumed to fall within the extent of the solar disc only. This paper is concerned with the second of these issues: examining the amount of energy that is available in a given radial displacement about the solar vector, that has been
*Corresponding author. Tel.: 161-2-9351-5979; fax: 161-29451-7725. E-mail address: buie@physics.usyd.edu.au (D. Buie).

reected off a mirrored surface in a generic solar concentrating system. To estimate this, an accurate model of the radial energy distribution of the sun must be established. Originally, work conducted by the Lawrence Berkeley Laboratories (LBL) (Grether et al., 1975; Noring et al., 1991) in the late 1970s and early 1980s measured the degree to which sunshapes varied from location to location. Eleven sites were chosen across the United States exhibiting different atmospheric characteristics such as altitude, proximity to sources of large particulates, and humidity. Grether, Evans, Hunt and Wahlig collated over 200 000 individual solar proles into what was called the reduced database (RDB). Between them, they published over 15 articles relating to their methodology of collecting and the spectral nature and distribution of circumsolar radiation (a full reference list of their contribution is provided by Noring et al., 1991). This work was an important step toward the creation of a terrestrial sunshape model. Work carried out by Rabl and Bendt (1982) in dening a sunshape using the RDB was followed by Schubnell et al. (1991), Schubnell (1992a,b), Steinfeld and Schubnell (1993), Neumann and Schubnell (1992), Neumann and Groer (1996), Neumann et al. (1998) and Neumann and

0038-092X / 03 / $ see front matter 2003 Elsevier Ltd. All rights reserved. doi:10.1016 / S0038-092X(03)00156-7

418

D. Buie et al. / Solar Energy 74 (2003) 417427

Nomenclature

x Fcs Fi u ud uD uL f k g L P s

Circumsolar ratio Energy contained within the circumsolar region of the terrestrial solar beam Total terrestrial solar beam Angular displacement from the solar vector Radial displacement of the solar disk Radial displacement of the circumsolar region Radial displacement dening the size of the solar cone Radial solar energy prole (sunshape) per steradian Sunshape variable Sunshape variable Percentage of the terrestrial solar beam Probability distribution of the combined surface error Standard deviation of the probability distribution

Witzke (1999). All of these authors made contributions to the description of the radial energy distribution of the sun and the effect this distribution has on the ultimate ux distribution in the focal region of concentrating systems. Neumann et al. (2002) described six sunshape proles that are indicative of a typical range of atmospheric conditions. These proles, combined with their statistical weights, represent a numerical database for calculating the inuence of variable conditions of the sunlight scattering on solar concentrating systems. Buie et al. (2003) extended the analysis of the LBL sunshape data, describing an empirical sunshape model that illustrates little variation over a range of geographic locations and presenting an algorithm that can be used to recreate sunshapes for simulating solar concentrators. In summary, work described in the literature to date involves simulation of the radial energy distribution of the sun for clear skies showing good agreement with a vast amount of observed data. Concentrating systems have the added complexity of reecting sunlight off at least one mirrored surface. The reection causes a broadening of the solar beam. The broadening is due to both the dispersion effects of nonspecular surfaces, and surface normal deviations from mirror shape errors. Although tracking errors are equally important for estimating the size of the ux distribution in the imaging plane of a solar concentrator, they do not broaden the solar image and therefore are excluded from these calculations. Johnston (1995) showed that these two sources of errors, both surface normal deviation and dispersion, could be combined for a mirrored surface, and the broadening of the solar image could be simulated by considering a probability distribution of errors. Therefore, if sufcient statistical information is gathered characterising mirror surfaces, the typical reected solar distribution off that surface is predictable. This paper denes a solar cone whose principle axis is aligned with the solar vector and that has a radial angular

displacement containing a pre-dened proportion of the terrestrial solar radiation. The effective size of the solar cone is then described for a range of solar conditions. The broadened distribution, or solar image caused by the reection of the solar cone off a mirrored plane are calculated. Clear trends are presented that show the dependence of the effective size of the solar image on the accuracy of a mirrored surface for different sunshapes. The effective size of the solar image at the absorber plane that must be accommodated in the design and optimisation of solar concentrating systems is then discussed.

2. The sunshape Buie et al. (2003) showed that terrestrial sunshapes show little variation from location to location when they are described in terms of their circumsolar ratio (CSR). The CSR ( x ) is dened as the radiant solar ux contained within the circumsolar region of the sky Fcs , divided by the incident radiant ux from the direct beam and circumsolar regions combined Fi ,

Fcs x 5] . Fi

(1)

If u is the radial displacement about the solar vector, the angular extent of the circumsolar region (uD ) is set to be 43.6 milliradians (mrad), complementary to an active cavity radiometer having an acceptance half angle of 2.58. The inner limit of the circumsolar region is the edge of the solar disc ud and has a generally accepted value of 4.65 mrad (0.2668) as calculated by Puliaev et al. (2000) at the Observatorio Nacional in Brazil. Since we are considering concentrating collectors only and therefore exclude insolation falling outside the circumsolar region, for brevity we will refer to Fi as the incident radiation. The incident radiation and the radiation contained in the circumsolar

D. Buie et al. / Solar Energy 74 (2003) 417427

419

region can be well-approximated by considering the integral of the sunshape f (u ) according to,

Fi,cs 5 2p

E f(u ) sin u du ,

uD

(2)

0,ud

2p

E f(u )u du ,
0,ud

uD

(3)

for small u. If the sunshape is known accurately, then the angular distribution of the incident radiation is well dened. The sunshape described by Buie et al. (2003), cos(0.326u ) ]]] for hu [ Ru0 # u # 4.65 mradj, f (u ) 5 cos(0.308u ) k g e u for hu [ Ruu . 4.65 mradj,

and that uses only one parameter, the CSR, based on the LBL reduced database and Neumann et al. (2002) sunshape proles. Typical CSR values vary from zero to 0.3, but have been recorded close to unity (Noring et al., 1991). Of the solar proles collected by Neumann et al. (2002), 71.5% had a CSR less than 0.05 and 89.5% of the sunshapes had a CSR less than 0.15; a complete statistical breakdown is presented in their work. Using a charged coupled device camera, 63 normalised sunshapes with a CSR of or about 5% were correlated by Neumann et al. (2002) from three sites across Europe. The comparison between this data and sunshapes created using Eq. (4) with CSRs of 4%, 5% and 6% can be seen in Fig. 1. Clearly, Eq. (4) is a good representation of independently observed data. Using Eqs. (16), the proportion L of the incident radiation bounded by a specic radial displacement uL about the central solar vector can be estimated from,

(4) where k and g are given by,

Ef(u,x)u du
L( x ) 5 ]]]] , uD
0

uL

(7)

k 5 0.9 ln(13.5x )x 20.3 , g 5 2.2 ln(0.52x )x 20.43 2 0.1 ,

Ef(u,x)u du
0

(5) (6)

represents a sunshape model that is independent of location

for any given CSR. Using Eq. (7) the effective size of the solar cone was estimated for solar radiation with CSRs ranging from 0.01

Fig. 1. Comparison between Eq. (4) and observed data collated by Neumann et al. (2002).

420

D. Buie et al. / Solar Energy 74 (2003) 417427

Fig. 2. The effective size of a solar cone containing various percentages of the incident radiation.

to 0.8 (Fig. 2). The six curves represent various percentages of the incident radiation of 70%, 80%, 85%, 90%, 95% and 98%. Obviously the upper asymptotic limit to the effective size of the solar cone is 43.6 mrad, encompassing 100% of the incident radiation, as dened here. By denition, a solar prole with a CSR of 0.3 has 70% of its energy contained within the solar disc. For each of the CSR sunshapes created using Eq. (4) the effective edge of the solar disc was determined (Fig. 3). The value of the angular displacement for all of the simulations should be 4.65 mrad, the accepted limit of the solar disc. This implies that the error associated with the prediction of the size of the solar disc is 61% for sunshapes with a CSR of less than 0.2 and 64% for the remainder of the sunshapes.

Johnston (1995) showed that by considering as a systemic mass both the slope error (dened as the angular deviation of the actual surface normal vectors from their ideal directions, measured in milliradians (Johnston, 1998)) and surface dispersion effects, the combined error could be treated by the probability distribution (P), dP u 2 2 ] 5 ] e 2u / 2 s , du s 2 (8)

2.1. Reection of a mirrored surface


If the terrestrial solar image is reected off a perfect (planar) surface, the reected image would be identical to the original. In real systems though, the reection of an image off a surface causes a distortion in that image. Two principle effects cause this distortion in solar concentrators. Firstly, real surfaces interact with the reected radiation, causing the expected specular reection to form a dispersive cloud. Secondly, mirror shapes are not perfect, that is to say, the variations in the surface normals from the ideal mirror shape are an additional inuence on the reected image.

where s is the standard deviation of the combined surface slope and dispersion error, and u is the radial displacement of a reected beam from the specular direction. A real surface can then be characterised by the standard deviation of the probability distribution. As an example of different surfaces, a high quality optical mirror has a probability distribution with a standard deviation of about 0.2 mrad, whereas poorer quality solar reectors could have a standard deviation as high as 8 mrad. To investigate the reection of the solar image off a non-ideal mirrored surface, the convolution of the solar image created from Eqs. (3)(6) and the probability distribution from Eq. (8) was conducted (Rabl, 1985, pp. 202205). The convolution in principal involves redistributing the vectors representing the solar image according to a normalised two-dimensional representation of a radial Gaussian distribution. Initially a solar image consisting of one hundred thousand appropriately weighted vectors

D. Buie et al. / Solar Energy 74 (2003) 417427

421

Fig. 3. Demonstration of the accuracy of Eq. (4) at predicting the edge of the solar disc for a range of atmospheric conditions. The angular displacement of the solar disc should be 4.65 mrad.

Fig. 4. The effective size of the solar cone after reection off a surface with a standard deviation of surface errors of 3.5 mrad.

422

D. Buie et al. / Solar Energy 74 (2003) 417427

Fig. 5. The angular displacement of the reected solar cone for different mirror errors for a CSR of 0.05. Curves are shown for the collection limits of different percentages of the incident radiation.

Fig. 6. The angular displacement of the reected solar cone for different mirror errors for a CSR of 0.1. Curves are shown for the collection limits of different percentages of the incident radiation.

D. Buie et al. / Solar Energy 74 (2003) 417427

423

Fig. 7. The angular displacement of the reected solar cone for different mirror errors for a CSR of 0.2. Curves are shown for the collection limits of different percentages of the incident radiation.

Fig. 8. The angular displacement of the reected solar cone for different mirror errors for a CSR of 0.3. Curves are shown for the collection limits of different percentages of the incident radiation.

424

D. Buie et al. / Solar Energy 74 (2003) 417427

Fig. 9. The angular displacement of the reected solar cone as a function of the circumsolar ratio for different mirror tolerances for an incident radiation collection efciency of 80%.

Fig. 10. The angular displacement of the reected solar cone as a function of the circumsolar ratio for different mirror tolerances for an incident radiation collection efciency of 90%.

D. Buie et al. / Solar Energy 74 (2003) 417427

425

Fig. 11. The angular displacement of the reected solar cone as a function of the circumsolar ratio for different mirror tolerances for an incident radiation collection efciency of 95%.

extending over the angular radius of the circumsolar region was created. A normalised two dimensional representation of a radial Gaussian distribution was also created, extending to an angular displacement of three standard deviations from each solar vector. The weight of each of the solar vectors was then redistributed according to the normalised Gaussian about the position of each vector, simulating a real surface with a standard slope error of s. It must be pointed out that the convolution of a radial error distribution and a solar image is only truly representative of a reected beam at normal incidence. For non-normal reection, being the majority of reections, the error distribution would be asymmetric (or elliptical). Therefore this method is only an approximation of a real system. The energy distribution of various solar images was calculated including the effective size of the solar cone, after the reection off assumed planar mirrors of differing optical qualities. Fig. 4 illustrates a typical result similar to Fig. 2, except that the original solar image was reected off a surface with a standard deviation of 3.5 mrad. The results of the simulations have been presented in two ways: rstly, the effective size of the solar cone was compared primarily to the optical quality of mirrors for particular solar conditions (Figs. 58) and secondly, the size of the solar image was primarily compared to various solar

conditions where the concentrator has certain predened parameters (Figs. 911).

3. Discussion Figs. 58 illustrate the dependence of the effective size of the reected solar cone on the quality of the reecting surface, for a generic solar concentrating system. As in Figs. 2 and 4, a range of incident radiation proportions (L) was chosen so as to get a clear indication of the optical performance of concentrating systems. Of the four solar conditions ranging from a CSR of 0.05 to 0.3, the effective size of the reected solar cone increases with the degradation of the optical quality of the mirrored surface (dened by the standard deviation of the combined surface errors). Also as the incident beam becomes broader (dened by a larger CSR), the effective size off the solar cone increases substantially, particularly for small surface errors. An approximately linear relationship exists between the angular size of the reected solar cone and the standard deviation of errors, as the standard deviation of errors increases. This is qualitatively different to the upper asymptotic limit in the effective size of the solar cone that exists in reecting the solar image off an ideal surface. As

426

D. Buie et al. / Solar Energy 74 (2003) 417427

the standard deviation of the surface error increases, the inuence of the sunshape distribution diminishes, effectively reducing the sunshape distribution to a point source. Therefore, the resulting reected energy distribution is effectively described by a radial Gaussian distribution where uL ~ s. Figs. 911 illustrate the effective size of the solar cone for various solar conditions, described by an increasing CSR from 0.01 to 0.8. The higher the quality of the reecting surface, the smaller the size of the angular spread of the solar image; analogous to a smaller absorber aperture that would be required in a generic solar concentrator. For locations in the world where there is little atmospheric scattering, represented by a low CSR, the optical quality of the mirror surface becomes more critical. For environments where there is generally large atmospheric scattering (signied by average CSRs of between 0.2 and 0.3), very little improvement in the collection efciency can be achieved by using mirrors with an optical quality better than between 4.5 or even 3 mrad in the standard deviation of the surface errors. Ideally, the size of the absorber aperture in solar concentrating systems should be approximately the size of the reected solar image in the focal plane, plus an allowance to account for the typical tracking errors of the reector modules. If all the power of the reected beam was thought to be within the connes of the direct beam radiation (4.65 mrad about the solar vector) then an overestimation of the collector efciency would occur. For the design of solar concentrating systems at a particular site, the size of the reected solar image can be dened by two parameters: the CSR and the standard deviation of the surface error of the mirrors. If enough sunshape information about a site is known then average, typical, or effective maximum CSRs may be predicted and used as a design parameter (see Rabl and Bendt, 1982). The solar distribution chosen for the design process may be related to improvements in the seasonal or daily output characteristics. The size of the absorber can then be determined by accurate knowledge of the standard deviation of the surface errors using Figs. 411. Alternatively, given an optimum absorber aperture and therefore acceptance angle, the quality of the reective surfaces can be chosen to ensure that a certain percentage of the incident radiation strikes the absorber. The absorber aperture may be determined by thermal considerations leading to a trade-off between optical efciency and the total conversion efciency of the concentrating system (Steinfeld and Schubnell, 1993). For concentrating photovoltaic systems, the sunshape spectral dependence could also be incorporated with a similar method, as previously described by Schubnell (1992b). Whilst the current paper does not outline an analytical design procedure for incorporating CSR considerations into solar concentrators, as for example given by Rabl and

Bendt (1982), the general relationship between the circumsolar ratio, mirror accuracy, absorber acceptance angle and optical efciency has been established in design curves covering a wide range of the important parameters. Further work will examine commercial realities such as the efciency of conversion devices, their costs, the non-linear dependence of mirror accuracy and mirror cost, in addition to the parameters investigated here.

4. Conclusion The inuence of sunshape, described by the circumsolar ratio, and mirror tolerance, described by the standard deviation of the mirror errors incorporating non-specularity and non-idea mirror shapes, on the angular distribution in the focal region of a generic solar concentrating system has been investigated. The results, obtained by detailed raytracing, and presented as a series of graphs covering the parameters over their typical ranges, allow broad design considerations to be made in the design of solar concentrating systems.

References
Buie, D., Monger, A.G., Dey, C.J., 2003. Sunshape distributions for terrestrial solar simulations. Solar Energy 74 (2), 113122. Grether, D.F., Evans, D., Hunt, A.J., Wahlig, M., 1975. Measurement of circumsolar radiation. In: Proceedings of Symposium of Optics in Solar Energy Utilization, 19th Annual Technical Symposium of The Society of Photo-Optical Instrumentation Engineers. San Diego, CA, pp. 1822. Johnston, G., 1995. On the analysis of surface error distributions on concentrated solar collectors. J. Solar Energy Eng. 117, 294296. Johnston, G., 1998. Focal region measurements of the 20 m 2 tilted dish at the Australian National University. Solar Energy 63, 117124. Mie, G., 1908. Beitrage zur optik truber medien, speziell kolliodaler metallo sungen. Ann. Physik 25, 377445. Neumann, A., Groer, U., 1996. Experimenting with concentrated sunlight using the DLR solar furnace. Solar Energy 58, 181 190. Neumann, A., Schubnell, M., 1992. Irradiance and sunshape measurements for the Cologne site. In: Proceedings of the 8th International Solar Forum, pp. 11731183. Neumann, A., von der Au, B., Heller, P., 1998. Measurements of circumsolar radiation at the Plataforma Solar (Spain) and at DLR (Germany). In: Proceeding of the International Solar Energy Conference Solar Engineering, pp. 429438, ISBN 0-7918-1856-X. Neumann, A., Witzke, A., 1999. The inuence of sunshape on the DLR solar furnace beam. Solar Energy 66 (6), 447457. Neumann, A., Witzke, A., Jones, S.A., Schmitt, G., 2002. Representative terrestrial solar brightness proles. J. Solar Energy Eng. 124, 198204.

D. Buie et al. / Solar Energy 74 (2003) 417427 Noring, J.E., Grether, D.F., Hunt, A.J., 1991. Circumsolar radiation data: the Lawrence Berkeley Laboratory reduced data base. In: National Renewable Energy Laboratory. NREL / TP-2624429. http: / / rredc.nrel.gov / solar / pubs / circumsolar / title.html Puliaev, S., Penna, J.L., Jilinski, E.G., Andrei, A.H., 2000. Solar diameter observations at Observatorio Nacional in 19981999. Astronomy & Astrophysics Supplement Series 143, 265267. Rabl, A., 1985. Active Solar Collectors and Their Applications. Oxford University Press. Rabl, A., Bendt, P., 1982. Effect of circumsolar radiation on performance of focusing collectors. J. Solar Energy Eng. 104, 237250.

427

Schubnell, M., 1992a. Inuence of circumsolar radiation on aperture, operating temperature and efciency of a solar cavity receiver. Solar Energy Mater. Solar Cells 27, 233242. Schubnell, M., 1992b. Sunshape and its inuence on the ux distribution in imaging solar concentrators. J. Solar Energy Eng. 114, 260267. Schubnell, M., Keller, J., Imhof, A., 1991. Flux density distribution in the focal region of a solar concentrator system. J. Solar Energy Eng. 113, 112116. Steinfeld, A., Schubnell, M., 1993. Optimum aperture size and operating temperature of a solar cavity-receiver. Solar Energy 50 (1), 1925.

Das könnte Ihnen auch gefallen