Evaluation of 12 Transposition Models Using Observations of Solar Radiation and Power Generation

Abstract

This article evaluates the performance of 12 solar transposition models according to the measured data in Zhangbei of Northern China, including four isotropic models, Liu&Jordan, Koronakis, Badescu, and Tian, and eight anisotropic models, Willmot, Bugler, Hay, Skartveit−Olseth, Stevenand−Unsworth, Reindl, Temps−Coulson, and Klucher. We measured global horizontal irradiance, beam global horizontal irradiance, and diffuse horizontal irradiance. Open circuit voltage and short circuit current data of tilted photovoltaic panels are also surveyed. The radiation data of tilted surface estimated by the transposition models are fitted with the power generation data calculated by open circuit voltage and short circuit current. The different transposition models are evaluated by three error indicators, which are, respectively, root mean square error RMBE, and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</i> <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> . The relationship between radiation and power generation is explored. In Zhangbei, the best performing anisotropic model is Reindl model. The unit area power generation of PV panels in this region is about 0.03−0.05 times of the value of unit area received radiation. We compare measured power generation with power generation estimated by the Reindl model. For most of data, the absolute errors of power generation are less than 0.1 kWh/m <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> , and relative errors of power generation are not more than 20%.