Dust soiling effects on decentralized solar in West Africa

West Africa's proximity to the Sahara Desert can cause photovoltaic (PV) systems to experience significant power generation reductions when particulates settle onto solar panel surfaces in a phenomenon known as soiling. Previous studies have created models to estimate PV soiling globally, but these models have several limitations in the region due to the accuracy of the input particular matter (PM) concentration values, the accuracy of methods for computing power loss, and the regional use of decentralized solar systems. Here we develop and apply a methodology for the development of a soiling estimation model based on the Coello framework, and evaluate it when using input PM concentration values from both a reanalysis dataset (MERRA-2) and a dataset derived from satellite-observations (WUSTL). This model reduces error in soiling-related power loss estimates by approximately 50% when compared to the unadjusted Coello approach. Using this model, we find soiling losses can exceed 50% in the most greatly affected locations during the Harmattan season and these power losses can vary considerably each year (between 26 and 53% in Niamey, Niger). To support PV system design, we develop a function that relates maximum soiling losses to annual cleaning frequencies in major cities. Finally, we find if decentralized solar is used to meet electricity access goals as projected, PV cleaning costs could sum up to $1Bn (USD) annually in sub-Saharan Africa by 2030.