Field spectra binning for energy production calculations and multijunction solar cell design

Annual spectra sets must be used for accurate energy production prediction and multijunction solar cell design for maximum energy production at a specific site. These spectra sets contain a large quantity of data that is cumbersome to manage during solar cell design calculations and impractical to reproduce in solar simulators for indoor energy production measurements. However, it should be possible to bin together spectra with similar spectral contents, and then use this reduced set with little loss of accuracy. We present two binning algorithms which judiciously bin together similar spectra to create a much smaller “proxy” set, for which the total measurement time, energy production calculation and solar cell optimization decreases to a matter of seconds. These algorithms are assessed against their accuracy in representing the whole spectra sets for solar cell design and energy production prediction. We find that a set of just five spectra fulfills this requirement. In addition, the sets of proxy spectra act as “fingerprints” of specific sites, and provide an efficient and effective way to understand how cell design and performance vary from site to site. Furthermore, the process of reducing a full data set to a few proxy spectra can help assess the quality of the dataset for multijunction applications, and contribute to improvements to the datasets and data collection methods.