GIS-based optimization of photovoltaic panel orientation in Benin using genetic algorithms and K-means clustering

Abstract

Accurately determining the orientation of fixed-tilt photovoltaic (PV) panels is critical for maximizing annual energy output, yet most installations in Benin continue to rely on empirical or generalized guidelines. This study develops a high-resolution (0.05° × 0.05°) geospatial database of optimal tilt (β*) and azimuth (α*) angles across the country by coupling the Perrin-de-Brichambaut clear-sky model with a Genetic Algorithm, that uses 2023 NASA-POWER meteorological inputs. Results show that β* increases from approximately 6° along the southern coast to around 13° in the far north, while α* evolves from a slight westerly orientation to a moderate easterly preference above 8°N latitude. To facilitate practical implementation, K-means clustering was applied separately to both angle datasets, with the number of clusters determined via the Kneedle algorithm. The resulting 3 tilt and 3 azimuth clusters combine into 8 implementation-ready mounting configurations that retain >95 % of the energy yield provided by fully site-specific optimization, while substantially simplifying system design. The accompanying orientation atlas, GIS layers, and open-source Python code offer a scalable and evidence-based planning tool for PV deployment in Benin, supporting policymakers, engineers, and developers in accelerating the energy transition.