Proceedings Article | 21 August 2009
KEYWORDS: Solar cells, Aerosols, Brain-machine interfaces, Atmospheric modeling, Atmospheric optics, Atmospheric particles, Photovoltaics, Sun, Quantum efficiency, Data modeling
This contribution addresses the need for more information about the spectral effect affecting solar cells specifically
designed for concentrating photovoltaic (CPV) applications. Spectral effects result from differences between the actual
(dynamically variable) solar spectrum incident on a solar cell in the field and the standard (fixed) solar spectrum used for
rating purposes. A methodology is proposed to quantify this spectral effect at any site where basic atmospheric
information exists, and predict what semiconductor material(s) may benefit from operating under non-standard
conditions. Using the same SMARTS radiative code as for the development of the improved reference spectrum for
concentrating PV rating, an analysis of the spectral sensitivity of five specific PV technologies to varying atmospheric
factors is presented, using simulated spectra at 5-nm resolution. (The alternative of using the average photon energy
(APE) concept was also considered, but proved inappropriate in the present context.) The technologies investigated here
include a 21.5%-efficient CIGS cell, a 22%-efficient crystalline silicon cell (both appropriate for low-concentration
applications), as well as three high-performance multijunction cells, which are specifically designed for high-concentration
applications. To the difference of most previous studies, the approach taken here considers realistic
atmospheric conditions. The proposed Daily Spectral Enhancement Factor (DSEF) is obtained from a typical daily-average
incident spectrum, which is purposefully weighted to minimize the incidence of large spectral effects at low sun.
Calculations of DSEF are performed here at fifteen world sites from an atmospheric monitoring network. These sites
have largely different latitudes and climates, and yet are all potentially interesting for CPV applications. Results are
obtained for a typical clear day of January and July, and for each of the five PV technologies just mentioned. This
analysis provides a preliminary quantitative assessment of how local atmospheric conditions interact with the spectral
response of different CPV technologies. Most importantly, it is shown that the effect of aerosol optical depth (AOD, also
referred to as atmospheric turbidity) has the largest impact on both the average direct normal irradiance (DNI) during a
given month and the cell's DSEF. It is found that DSEF can be as low as 0.993 under clean conditions (low AOD), and
as high as 1.215 under hazy conditions (high AOD). Under most conditions, all simulated solar cells perform
significantly better than under rating conditions due to the spectral effect alone. There is no important difference in
DSEF from cell to cell, except in one instance of very high AOD. The methodology and results proposed here constitute
a step towards a better performance prediction of CPV systems, by assessing the variable spectral effect more accurately.
It is anticipated that a more detailed simulation, which would also model temperature effects, as well as current-limiting
effects in multijunction cells, would indicate even larger DSEF values than found here. Accurate aerosol data with
higher spatial resolution in the "sun belt" than what exists today would also be desirable for the development of CPV
applications.
