A Data-Driven Methodology for Precision Flux Density Predictions for Heliostat Fields

Kurzfassung

The imperative transition to renewable energy sources necessitates advancements in technologies such as Concentrated Solar Thermal (CST) systems. A significant challenge in CST is the accurate characterization of heliostat beams, traditionally reliant on detailed surface deformation measurements—a process that is both costly and time-intensive. This dissertation introduces an innovative data-driven methodology that bypasses the need for such extensive physical modeling. Central to this approach is a generative neural network framework trained directly on focal spot data. By abstracting heliostat properties and learning the variations in focal spots under different conditions—such as sun angles and aimpoints—the model predicts flux distributions with high accuracy. This unified predictive pipeline aggregates data from the entire heliostat field, capturing shared patterns while adapting to individual heliostat variations. Validation using real-world data from the Solar Tower Jülich demonstrated the model’s robustness. The data-driven pipeline achieved a prediction error of 11% for individual focal spots, surpassing current state-of-the-art methods that rely on extensive measurements, in both robustness and scalability. When considering factors such as tracking errors, Direct Normal Irradiance (DNI), and reflectivity uncertainties, the total flux prediction error, aggregated from all heliostats, was approximately 5%. System-level analysis revealed that the enhanced accuracy of focal spot predictions translates into significant efficiency gains. The data-driven pipeline demonstrated efficiency improvements of over 4–16%, depending on receiver complexity. In summary, this research presents a scalable, cost-effective, and highly accurate framework for heliostat flux prediction. By eliminating the dependency on extensive physical measurements, it offers a practical alternative for CST systems, paving the way for more efficient and sustainable solar energy technologies.