Validation of the PVLife model using 3 million module-years of live site data

Establishing a strong basis for confidence in a solar technology requires being able to prove a low-degradation track record in the real world, and rationalize it with strong physical understanding and investigation. This paper briefly reviews our previously-published physical model for calculating degradation and reliability, PVLife, which computes hour-by-hour degradation of PV modules using weather files and degradation sub-models developed from accelerated test data. We then demonstrate a validation of this model against a large statistical data set obtained from 266 systems powered by SunPower modules (data from over 179 systems installed by Powerlight, using non-SunPower modules are also shown). In total these data represent over 800,000 modules and a total of 3.2 million module-years of experience. The data analysis technique requires little manual data processing and can be derived from live sites without special experimental treatment. We also discuss returnrate data on modules incorporating SunPower's back-contact cell, as well as front contact modules in SunPower's fleet. Implications for failure prediction are discussed.