Fast annual energy estimation for solar tower power systems based on heliostat grouping and solar position sampling

A three-level acceleration method to significantly improve the efficiency of annual energy estimation for solar power tower systems is proposed in this study. The first level reduces memory consumption and accelerates computation by discretizing the receiver surface into strips in Monte Carlo ray tracing flux spot simulation. The second level estimates the instantaneous optical efficiency of the heliostat field by grouping heliostats and simulating only one representative heliostat per group. Three grouping methods are explored: Recursive Binary Tree Grouping, K-means Grouping, and Uniform Grouping. The third level reduces the number of required simulation time points by sampling solar positions in the equatorial coordinate system and exploring B-spline or Radial Basis Functions to approximate the instantaneous optical efficiency. The proposed method is demonstrated using PS10 and Gemasolar heliostat field cases. Results show that the proposed method estimates the annual energy of the PS10 field in 2.921 s with a 329 times speedup within a 0.019% annual energy power error. For the Gemasolar field, it achieves a 1283 times speedup in 5.153 s with a 0.050% annual energy power error, demonstrating its potential for heliostat layout optimization and related applications.