Switzerland aims to install 30 to 40 GWp of photovoltaic capacity in the next 30 years to meet Energy Strategy 2050 targets. This goal is complicated by the prevalence of small- and medium-scale residential and commercial systems, which face higher risks of degradation and have limited data quality compared to larger utility-scale systems. This work aims to evaluate the performance of PV systems in the built environment in Switzerland through two lenses: (1) identifying reversible and non-reversible loss effects and their impact on system health, and (2) developing new data-based methodologies to determine PV system long-term performance with minimal uncertainty. This process aims to optimise PV system design, thereby increasing energy yield and helping to meet Switzerland's 2050 energy targets. A fault detection and diagnosis algorithm (FDDA) was developed and validated using the acquired datasets, successfully identify partial shading, snow and downtime losses. In parallel, an alternative to the most common performance loss rate statistical method (the year-on-year (YoY)) was developed, the multi-annual year-on-year (multi-YoY), which improves both accuracy and precision in long-term performance assessments.
Main findings («Take-Home Messages»)
- A large representative dataset of BAPV and BIPV systems in Switzerland was collected and analysed, focusing on improving our understanding and optimising PV system performance in the built environment. The project successfully identified key loss patterns and their effect on system reliability, highlighting the importance of planning and maintenance to reach the Swiss 2050 Energy Strategy targets.
- The combination of fault detection methods with long-term performance data analytics brings many interesting insights regarding the reliability of PV systems. The double approach enables the decoupling of intrinsic and extrinsic performance loss effects. The results highlight the importance of considering loss factors when analysing PV system data, as they can cause significant biases in the final PLR value.
- The multi-YoY approach, proposed as an alternative to the standard YoY method for evaluating PLRs, demonstrates enhanced accuracy and precision, particularly effective for linear performance loss trends. The primary benefit of the multi-YoY method lies in its ability to minimize uncertainty, achieving significantly narrower confidence intervals (reduced by one order of magnitude) compared to the traditional YoY method.
This work establishes the tools and methodology needed to streamline the analysis of long-term PV plant performance with increased accuracy and precision. These tools enhance the statistical significance of such analyses, as well as the general understanding of loss mechanisms in PV systems, creating a positive feedback loop that improves the design phase for the next generation of PV systems.
