Application of Dynamic Multi-Step Performance Loss Algorithm

In this work we present the application of a novel multi-step performance loss algorithm for PV systems and its practical implications. We try to better understand performance impairing effects and common occurrence patterns. The algorithm automatically detects the amount and position of breakpoints in non-linear performance ratio time series of PV systems, divides the time series into sub-parts and evaluates them independently in a linear fashion. Afterwards, based on the position of the breakpoints, system inspection data are used to find root causes for related changes in performance. The methodology is applied to two systems with an overall linear performance loss clearly below 1%/a. The results show similar performance loss patterns. The systems are two medium sized non-residential plants. After applying tailored data filter both systems experience a nearly linear performance loss over time with a slight performance fluctuation at the beginning of operation. This work focuses on system data and issues on system level. The study of the breakpoint-root causes correlation highlighted the necessity of tailoring the initial data filtering to the final objective of the analysis. It has been shown, that the presented algorithm simplifies the detection of performance instances which deviate from the norm.