Photovoltaic production in West Africa: Impact of dust and water footprint of cleaning operations

Abstract

To achieve universal electricity access and comply with Paris Agreement, one large-scale objective of the Economic Community of West African States (ECOWAS) is the deployment of +8 to +20 GWp of solar energy systems by 2030 (IRENA, 2018). ECOWAS is located south of the Saharan region and close to the Bodélé depression, which has been observed to have the largest atmospheric dust production activity on Earth (Isaacs et al., 2023). Once deposited on panels, dust reduces the transmission of solar radiation to the panels and, consequently, the energy production (Sarver et al., 2013). Annual losses of solar energy production of up to 54% have been observed in the region due to dust (Chanchangi et al., 2022). These production losses can be mitigated by regularly cleaning solar panels. In West Africa, cleaning operations commonly use water but many areas are water-scarce. It is thus important to ensure that water resources are not further strained by water cleaning operations associated with the expected large-scale deployment of solar energy systems in the region.In the present work, we aim to assess the water footprint of different cleaning strategies of virtual solar plants in the ECOWAS region. A first step towards this aim consists in regionally assessing how dust would accumulate on Photovoltaic (PV) panels and, in turn, what the associated production losses would be. We present a dust accumulation model allowing to simulate, over a long time period and across the region, the temporal sub daily variations of dust accumulation on virtual PV panels. The model uses as input the particulate matter concentration of different particle sizes. Dust data from the CAMS and MERRA2 reanalyses are considered. Both datasets are first compared to observations of regional particulate matter concentration available from a set of four stations from the INDAAF network. CAMS data were found to better agree with observations (> 0.8 correlation for a 1-week temp