An investigation of the key parameters for predicting PV soiling losses

Abstract

One hundred and two environmental and meteorological parameters have been investigated and compared with the performance of 20 soiling stations installed in the USA, in order to determine their ability to predict the soiling losses occurring on PV systems. The results of this investigation showed that the annual average of the daily mean particulate matter values recorded by monitoring stations deployed near the PV systems are the best soiling predictors, with coefficients of determination (R 2 ) as high as 0.82. The precipitation pattern was also found to be relevant: among the different meteorological parameters, the average length of dry periods had the best correlation with the soiling ratio. A preliminary investigation of two-variable regressions was attempted and resulted in an adjusted R 2 of 0.90 when a combination of PM 2.5 and a binary classification for the average length of the dry period was introduced.

Predicting photovoltaic soiling losses using environmental parameters: An update

“…The soiling station data have been analyzed using the method described in Micheli and Muller . Soiling has been quantified using the soiling ratio (r s,i ), which is a metric expressing the ratio, in percentage, between the short‐circuit current of the soiled device and the short‐circuit current of the control device.…”

“…In the recent years, several studies have investigated the relations between soiling losses and environmental parameters. The analysis of data recorded by 20 soiling stations installed in the United States has shown that the average concentrations of particulate matter and the average length of the dry period at each sites were the best parameters to predict the average soiling occurring over the long periods of data collection . Recently, a study conducted on performance and environmental data collected in Doha, Qatar, has shown how the prediction of daily soiling losses relies on complex correlations among multiple variables .…”

“…The two studies do not contradict each other: it has already been discussed in literature how the prediction of short‐term soiling losses can be more difficult than the prediction of annualized losses because the variability of the environmental parameters is more relevant for short‐term analysis . Moreover, the scopes of the two works are different, as the modelling in Javed et al was performed to predict the daily soiling losses at a single site, whereas the analysis in Micheli and Muller was conducted to rank the yearly soiling losses at 20 sites through the analysis of a number of variables in order to understand if the differences in soiling losses among sites could be predicted through various environmental parameters.…”

“…The aim of the present paper is to extend the analysis presented in Micheli and Muller by using data from a larger number of soiling stations through both single‐variable and multi‐variable regressions. This analysis contributes to the effort being made to reduce the uncertainty in predicting soiling, to provide the community with alternative methods to estimate annualized soiling losses even when soiling data are not available.…”

“…Being able to estimate the average impact of soiling on the annual energy yield at a site before a PV system is built would lead to a better cost evaluation and site selection, and to the optimization of the system design to limit the accumulation of soiling. Compared with the previous work, the number of variables investigated has been increased as well, to (1) include new parameters that have been reported to impact the transmissivity of glasses, and (2) analyze how different resolutions, sources, or interpolation techniques can vary the outputs of the analysis.…”