EvoRec: Simulation and optimization of solar tower receivers based on annual performance assessment with ANN and evolutionary algorithms

In this study, an optimization approach for solar tower receivers is presented. At its core is a detailed optical and thermo-hydraulic simulation model, built on ray tracing and a spatially resolved heat and pressure loss model. This detailed physical model is incorporated in an approach for dynamic system simulation, that uses a sky discretization and flux level interpolation approach for fast optical assessment and accelerates the annual performance assessment by means of an artificial neural network. Using an objective function based on the introduced modeling approaches, the receiver is optimized for effective annual thermal gain. The entire optimization methodology is called EvoRec. The developed methodology is demonstrated for a reference system, which has been modeled based on data from literature to resemble the Gemasolar plant in Spain. By optimizing the receiver in terms of six degrees of freedom, a relative increase of annual yield by 12% compared to the reference setup is reached.