Crop-driven optimization of agrivoltaics using a digital-replica framework

Abstract

Agrivoltaics are a novel form of agricultural production where photovoltaic panels are blended with crops in order to optimize land use, particularly with respect to crop production and power generation. Given agrivoltaics are complicated systems where crop production, water use efficiency, land use efficiency, solar energy production, and the economics of the entire system are all dependent and competing for solar energy, there is opportunity to develop models incorporating these objectives into an optimizable framework. This work contributes to agrivoltaic design methodology through a digital replica and genomic optimization framework which simulates light rays in a procedurally generated agrivoltaic system at an hourly timestep for a defined crop, location and growing season to model light absorption by the photovoltaic panels and the crop below. Hourly radiation values are then summed into daily radiation values and fed into a crop model to simulate performance of an agrivoltaic and a reference crop at a daily timestep. The results of photovoltaic and crop performance metrics for a given design are then used in a genomic optimization algorithm to conduct a multi-objective optimization across various designs to find an optimal, crop-driven solution for a defined crop, season and location. A numerical example is demonstrated using this framework with a SunnySD tomato crop grown in Davis, California, resulting in 28.9% optimization of combined crop and energy production using a genomic optimization scheme over 50 generations.