Performance evaluation of radiation sensors for the solar energy sector

Rotating Shadowband Irradiometers (RSI) and SPN1 Sunshine Pyranometers allow determining the diffuse and direct components of solar radiation without sun trackers; they can be deployed in networks for continuous field operation with modest maintenance. Their performances are evaluated here by analyzing their errors with respect to well characterized references. The analysis is based on 1‑minute data recorded over a 15‑month period at the Payerne BSRN station in Switzerland. The analysis was applied both to the whole dataset and data subsets reflecting particular conditions to allow a better understanding of how instrument performance depends on such conditions.The overall performance for measuring global horizontal irradiance (GHI) is satisfactory with deviations compatible with an expanded uncertainty of ±25 Wm−2 (±10 %). For diffuse horizontal irradiance (DfHI), RSIs exhibited errors on the order of ±20 Wm−2 (±13 %) with some of them being affected by small systematic negative biases on the order of −5 Wm−2 (median). SPN1s underestimate DfHI by about −10 Wm−2 (median) with a relatively large range of the expanded error distribution between −45 Wm−2 and 20 Wm−2 (−35 % to 13 %). For direct normal irradiance (DNI), the extended error range for RSIs is on the order of ±40 Wm−2 (±5–6 %) with some instruments presenting no bias while others are affected by median biases up to −15 Wm−2. SPN1s exhibit a relatively large median bias of 40 Wm−2, and an extended range of the error distribution between −45 Wm−2 and 125 Wm−2 (−6 % to 19 %). Typical errors on the integrated yearly energy per unit surface area are on the order of a few percent or less (< 5 %) for RSI with negligible errors on DNI for some RSI instruments. SPN1 integrated errors are negligible for GHI, but on the order of −8 % for DfHI, and between 9 % and 11 % for DNI. For RSIs, GHI and DfHI errors showed similar amplitude and dependence on solar elevation, while DNI errors were significantly smaller in relative terms than GHI or DfHI errors. This suggests that RSIs are optimized for providing good estimates of DNI, at the expense of – and resulting in – a correlation between GHI and DfHI errors. RSI uncertainty for DNI is about twice the uncertainty of a good quality pyrheliometer under favorable conditions. SPN1 instruments exhibit the opposite behavior with GHI and DfHI errors of opposite signs, resulting in large DNI errors. While the SPN1 performances for measuring GHI are similar to those of RSI, corrections are required to obtain satisfactory performances for DNI.