On the Uncertainty of Estimating Photovoltaic Soiling Using Nearby Soiling Data

Leonardo Micheli, Matthew Muller

Research output: Contribution to journalArticlepeer-review

Abstract

The accumulation of dust on the surface of photovoltaic modules can reduce their performance and affect the cost competitiveness of this technology. This phenomenon is known as soiling and can be mitigated through appropriate corrective and/or preventive actions. In order to maximize its effectiveness, it is important to plan the soiling mitigation strategy even before the PV system is operational. This is typically done through a nearest neighbor approach, by estimating soiling using data from the nearest operational photovoltaic system. This work focuses on understanding the uncertainty related to this practice. For this purpose, the semi-variance function is used to study the dissimilarity between the soiling losses of two locations in California depending on their distance. The results show that, when the soiling loss at a nearby system is used to estimate soiling of a site, the uncertainty can be approximated to increase linearly at a rate of 0.08–0.10%/km up to 60 or 80 km. After this distance, the use of a nearest neighbor approach is no longer justified, as it produces an uncertainty as big as the average soiling loss of the sites in the dataset used in this study. In some conditions, uncertainties > 0% are found also for sites located within 25 km, meaning that even close-by systems might soil differently.

Original languageAmerican English
Article number100120
Number of pages7
Journale-Prime - Advances in Electrical Engineering, Electronics and Energy
Volume3
DOIs
StatePublished - Mar 2023

Bibliographical note

Publisher Copyright:
© 2023 The Author(s)

NREL Publication Number

  • NREL/JA-5K00-81014

Keywords

  • Nearest neighbor
  • Photovoltaic systems
  • Soiling
  • Solar energy
  • Spatial interpolation

Fingerprint

Dive into the research topics of 'On the Uncertainty of Estimating Photovoltaic Soiling Using Nearby Soiling Data'. Together they form a unique fingerprint.

Cite this