Performance Benchmark of Radiometers for Solar Applications under the Influence of Meteorological Parameters

Abstract

Accurate measurements of direct normal irradiance (DNI) and diffuse horizontal irradiance (DHI) are essential for the planning and operation of solar power plants as well as for climate monitoring. The use of Sun trackers in combination with pyranometers and pyrheliometers of the highest quality class according to ISO 9060, currently provides the most accurate data. At the same time, however, this measurement equipment is always associated with high acquisition and maintenance costs. In this thesis, the performance of a total of four radiometers for solar applications is evaluated and the influence of various atmospheric conditions on the measurement accuracy is estimated. The sensors investigated include a rotating shadowband pyranometer (RSP) equipped with an LI-COR 200R, an SPN1 from Delta-T Devices, the comparatively new MS-90 from EKO Instruments Co., Ltd., and the recently developed, camera-based, PyranoCam method from DLR. The study is carried out in a semi-desert climate in Tabernas, Spain. Performance is evaluated using the error metrics RMSD, MAD, BIAS, and the 95% prediction interval (PI95), representative for the measurement uncertainty. The evaluation of the dependence on meteorological influences is done mathematically using Spearman’s rank correlation coefficient as well as visually with the help of scatter density plots. For DNI performance under all-sky conditions, the PyranoCam method shows the best results with an RMSD of 3.4% and a MAD of 2.4%. This is followed by the RSP and the MS-90. SPN1 performs worst (RMSD=13.2%, MAD=10.4%). The narrowest PI95 can be found for the RSP (-39.2 to 11.4 W/m2), closely followed by PyranoCam. The widest PI95 is between -32 and 140.2 W/m2 for SPN1. For DHI measurement, RSP shows the smallest RMSD (4.7%) and MAD (3.6%). PyranoCam is in second place. MS-90 shows due to its high RMSD (22.9%) and MAD (14.5%) values the worst results. Here too, RSP provides the narrowest PI95 (16.7 to 3.2 W/m2), once more, followed by PyranoCam. The largest measurement uncertainty occurs when measuring DHI with the MS-90 (-71.4 to 56.2 W/m2). Regarding the atmospheric dependencies, the PyranoCam method is most strongly influenced by the hourly-averaged DNI clear sky index, which reflects the cloudiness level, in both the DNI and DHI measurements. However, since the cloudiness of the sky has a significant influence on the spectral composition of the radiation, it could be that the overestimation of PyranoCam’s DHI is presumably caused by a higher weighting of the solar spectrum in the red region. The RSP shows for both, the DNI and DHI measurement, a moderate dependence on the Sun’s elevation angle. While for DNI, the circumsolar radiation affects SPN1’s measurement the most, for DHI it is the DNI clear sky index representative for the effective wavelength, which marks the main influence. With respect to the MS-90, circumsolar radiation is the main influence here for the measurement of both DNI and DHI.