Analysis of Variability in Calculated Performance Loss Rates of Large-Scale PV Plants

The performance loss rate (PLR) is a key parameter in predicting and measuring the lifetime production from a PV plant. In this study, PLR is calculated by using historical on-site performance and meteorological time-series data from common, existing sensors and instrumentation at large-scale PV plants and the RdTools software package. Results show that current analysis techniques are not precise enough to identify performance improvement opportunities at any single plant but can enable insights at an aggregated fleet-level. The high variability in the results over time combined with the correlation among subarrays within a plant suggest that factors in addition to module degradation, such as soiling and un-normalized weather, have a significant impact. In addition, the reported confidence intervals are likely an underestimate of uncertainty due to the non-independence of daily calculated loss rates. This work examines PLR results to gain insight on how the analysis precision might be improved to better identify underperformance at the plant level and, subsequently, enable performance remediation opportunities. Also, a method to quantitatively validate loss rate analysis techniques using synthetically generated performance data is proposed.