Uncertainty estimation of temperature coefficient measurements of PV modules
Temperature coefficients of PV modules play an important role in distinguishing between products in an increasingly competitive market. However, measurement setups vary greatly and inter-laboratory comparisons show deviations from the mean of around ±10-15 %, or even larger, for temperature coeffici...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Default Article |
| Published: |
2016
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/2134/22306 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Temperature coefficients of PV modules play an important role in distinguishing between products in an increasingly competitive market. However, measurement setups vary greatly and inter-laboratory comparisons show deviations from the mean of around ±10-15 %, or even larger, for temperature coefficients of maximum power. Measurement deviations often do not agree with the uncertainty estimates indicating that uncertainty is significantly underestimated. On the other hand, some laboratories have adopted a very conservative approach and needlessly overestimate the uncertainty. A new and robust methodology for calculating the temperature coefficients is presented here. This includes estimating and propagating the uncertainty of different types of measurement systems and procedures, in accordance with international standards. The method is validated with a round-robin inter-comparison. Two c Si modules were measured with five different measurement setups with uncertainties estimated following the proposed approach. The advanced uncertainty estimation method resulted in a decrease of the estimated uncertainty of all systems by a minimum of 50 % compared to the previous conservative estimates, enabling us to identify a previously unknown systematic effect. The measurement results of one of the systems were inconsistent with the estimated uncertainty. Further investigation confirmed a systematic effect due to the poor spectrum of that system. Removing the outlier measurement, the measurement percentage deviation from the reference value for maximum power temperature coefficients was within ±3.2%. The deviation was consistent with the stated uncertainties. The approach can facilitate the reduction of temperature coefficient measurements uncertainty by highlighting areas of improvement for bespoke systems. |
|---|