Testing Global Models of Photovoltaic Soiling Ratios Against Field Test Data Worldwide

Models for predicting photovoltaic soiling ratios are evaluated against soiling values reported in the scientific literature for 20 locations worldwide. These models assume soiling increases linearly over time until the photovoltaic arrays are cleaned or sufficient rain falls. Such models have been used before. The novelty in our approach is in exploring how the soiling rates can be estimated anywhere in the world, and in testing the resulting estimates against values reported in the literature. Our results suggest that these models can correctly reproduce broad differences in soiling between large regions, but do not capture variations at smaller spatial scales.

I. Introduction

Soiling losses are a major concern for those who develop and finance large-scale photovoltaic (PV) systems. Uncertainty about expected soiling losses and about the required cleaning frequency and maintenance costs can directly impact a project’s finances. Moreover, soiling losses vary tremendously, from being completely negligible in some locations to wiping out 60%of production within one day in others [1].