Multi-Annual Year-on-Year: Minimising the Uncertainty in Photovoltaic System Performance Loss Rates

Abstract

The performance loss rate (PLR) is a key parameter in the assessment of photovoltaic (PV) systems' long-term performance and reliability. Despite the lack of industry-wide consensus and standardised methods for extracting PLR values from field data, the year-on-year (YoY) method is often considered the most robust regression analysis. However, achieving reproducible results with minimal uncertainty remains a challenge. This work proposes the multi-annual YoY (multi-YoY) approach, which reduces the statistical uncertainty of the metric through increased usage of available data. The concept is straightforward: Instead of comparing data points only to the following year, the multi-YoY method compares them to all subsequent years, increasing the number of available comparisons. The methodology is validated using synthetic data and tested on high-quality datasets made available by IEA PVPS Task 13. The multi-YoY method improves both accuracy and precision, with only 1% deviation from the set PLR value in a synthetic dataset and a tenfold decrease in confidence interval (CI) compared to the standard YoY. Moreover, comparisons with the IEA benchmark PLR values show good agreement with their ensemble method, with minimised uncertainty. The impact of noise, dataset length missing data and non-linear trends are tested, showing improved accuracy and robustness for the multi-YoY approach. For non-linearity, automatic segmentation is recommended to capture the evolving PLR.