Data Filtering Impact on PV Degradation Rates and Uncertainty (Poster)

To sustain the commercial success of photovoltaics (PV) it becomes vital to know how power output decreases with time. In order to predict power delivery, degradation rates must be determined accurately. Data filtering, any data treatment assessment of long-term field behavior, is discussed as part of a more comprehensive uncertainty analysis and can be one of the greatest sources of uncertaintyin long-term performance studies. Several distinct filtering methods such as outlier removal and inclusion of only sunny days on several different metrics such as PVUSA, performance ratio, DC power to plane-of-array irradiance ratio, uncorrected, and temperature-corrected were examined. PVUSA showed the highest sensitivity while temperature-corrected power over irradiance ratio was found to bethe least sensitive to data filtering conditions. Using this ratio it is demonstrated that quantification of degradation rates with a statistical accuracy of +/- 0.2%/year within 4 years of field data is possible on two crystalline silicon and two thin-film systems.