Remote Diagnosis of Solar Panel Performance: A Case Study of the Filibaba Energy Kiosk

This paper describes a statistics-based investigation of the long-term power production of solar panels installed in September 2015 in an energy kiosk in Filibaba, Zambia. The panels power a centralized energy kiosk, which in turn is used for powering personal electronic devices, homes, and businesses. As with any off-grid system, faults and normal or accelerated degradation can reduce the power produced by the solar panels over time. The motivation of this work is to determine if the condition of the solar panels can be assessed remotely, using limited measured data. This can save the time and expense of sending technicians to the site to diagnose or troubleshoot the system. The methods discussed in this paper utilize a data set of the solar panel production and energy kiosk battery voltage. The methods include: comparing solar power output on days that could be expected to have similar weather and sun exposure; analysis of the consistency of power output as it relates to time of year; and using the battery voltage to find times of day for full power output of the panels. Preliminary results show that for the Filibaba energy kiosk, there is, on average, as much as a 34% decrease in power output from one year to the next. Given the age of the panels, these results prompt more analysis to determine if this is indicative of accelerated degradation. The work outlined in this paper will be used to further develop systems for remotely diagnosing issues that may cause decreased panel performance.