Minimising the Levelised Cost of Electricity for Bifacial Solar Panel Arrays using Bayesian Optimisation

TL;DR

This paper introduces a Bayesian optimisation approach to minimize the levelised cost of electricity in bifacial solar panel arrays, outperforming traditional design guidelines and providing adaptable solutions for various locations.

Contribution

The study develops a detailed illumination model and applies Bayesian optimisation to identify cost-effective solar panel configurations, surpassing established design standards.

Findings

Up to 23% lower LCOE achieved with optimized configurations.

Bayesian optimisation effectively identifies global minima in complex parameter spaces.

Design guidelines based on the algorithm outperform traditional latitude-based tilt and spacing rules.

Abstract

Bifacial solar module technology is a quickly growing market in the photovoltaics (PV) sector. By utilising light impinging on both, front and back sides of the module, actual limitations of conventional monofacial solar modules can be overcome at almost no additional costs. Optimising large-scale bifacial solar power plants with regard to minimum levelised cost of electricity (LCOE), however, is challenging due to the vast amount of free parameters such as module inclination angle and distance, module and land costs, character of the surroundings, weather conditions and geographic position. We present a detailed illumination model for bifacial PV modules in a large PV field and calculate the annual energy yield exemplary for two locations with different climates. By applying the Bayesian optimisation algorithm we determine the global minimum of the LCOE for bifacial and monofacial PV fields at these two exemplary locations considering land costs in the model. We find that currently established design guidelines for mono- and bifacial solar farms often do not yield the minimum LCOE. Our algorithm finds solar panel configurations yielding up to 23 % lower LCOE compared to the established configuration with the module tilt angle equal to the latitude and the module distance chosen such that no mutual shading of neighboring solar panels occurs at winter solstice. Our algorithm enables the user to extract clear design guidelines for mono- and bifacial large-scale solar power plants for most regions on Earth and further accelerates the development of competitively viable photovoltaic systems.