Maps of long-term soiling losses in Europe considering a partial cleaning effect by rain

The soiling of PV modules has been estimated to yield global losses in the solar energy production by 4 to 7%, even considering the cleaning efforts in many solar energy systems. Despite this, soiling measurements are not always available, making models particularly valuable for estimating these losses. However, soiling models currently present two limitations: most of them assume that daily rain accumulations higher than a threshold value can totally clean PV modules and, additionally, that rain will have the same washing effect on all types of soiling, independently of their higher or lower adherent properties. In this work, the HSU model (Coello and Boyle, 2019) is modified to include the two aforementioned effects. The original and modified HSU models are then calibrated for two locations in Africa where observations of long-term soiling losses were available. At the first considered location, the predominant soiling type was mainly washable by rain but showed the partial cleaning effects, while at the second location, despite the frequent and abundant rainfall, the build-up of persistent soiling was observed. The calibrated parameters for the original and modified HSU models were then applied to reanalysis meteorological 2-dimensional input data to estimate the associated soiling losses in Europe for a 20-yr operation without any artificial cleaning. The application of model parameters derived for the African sites for Europe is considered to be valid for the demonstration of the method and delivers exemplary results that might also occur for soiling types found at some European sites. The derived different soiling maps should, however, not be understood as the most likely possible values for European soiling. When considering the exemplary removable soiling type, the original HSU model underestimates on average the European soiling losses by a factor of 4 compared to the estimations of the modified HSU model. When considering the exemplary persistent soiling type, the European soiling losses estimated with the modified HSU model are on average 5.5 times larger than those estimated with the original HSU model. Additionally, the European soiling losses estimated with the modified HSU model considering a persistent soiling type are 2.4 times larger than those considering a removable soiling type, which highlights not only the importance of properly modeling the soiling losses but also of choosing the correct soiling type. The results show that, at least for some soiling types, mechanical cleaning of the PV modules is necessary to avoid high soiling losses even in rainy regions such as Central Europe.