Evaluation of Methods for Optimizing a Ray Tracing Simulation with a Bifacial Vertical Fixed-Tilt Agri-PV System

This work deals with the application of methods from the Global Sensitivity Analysis framework on a PV simulation use case. Four methods (EET, FAST, RSA, PAWN) are used to analyze the influence of nine input factors on the output of a ray tracing simulation done with bifacial_radiance in the example of a bifacial vertical fixed-tilt agri-PV system. Albedo and row spacing are identified to have the highest influence. Furthermore, it is found that a system size of 10 rows and 7 modules per row meets a semi-finite assumption. These results are used to derive optimization measures for the ray tracing simulation. The results obtained are very consistent throughout all methods. Since RSA and PAWN allow for factor mapping it is recommended to use one of them in combination with a method based on a different sampling strategy for cross-validation. The general use of GSA methods in the context of PV modelling is emphasized since the methods apply to all kinds of simulation models.