Questioning the reliability of open-loop calibration methods: Introducing a robust data sampling for year-round high accuracy

Abstract

Heliostat calibration in solar tower plants is critical for optimizing plant efficiencies through precise solar tracking. Current practices often assume heliostat precision degrades over time, leading to the development of new calibration procedures utilizing time-dependent data sets. However, contrary to this prevailing assumption, our study demonstrates the consistency of tracking accuracy over extended periods when appropriate calibration points are selected. We introduce a novel data sampling method that uses sun positions in Euler angles as relevancy scores, enabling higher accuracy with a reduced data requirement. Our thorough analysis challenges the common belief that time significantly impacts calibration accuracy. Furthermore, we unveil an overlooked relationship between prediction accuracy and solar position coverage, raising legitimate concerns about the reliability of reported accuracies in previous publications. To promote transparency, we present clear data and advocate for improved reporting practices in future publications. Applying the new data set sampling to a non optimized data set we archive a year-round stable accuracy below 1.5 mrad with as little as 27 calibration points.