Objective: Photovoltaic (PV) power plants are relatively simple systems, yet high-quality performance characterization remains elusive. Summary metrics such as Performance Ratio contain too many confounding effects and are often plagued by data quality issues. Our objective is to introduce a comprehensive characterization framework that encompasses data capability, event management and operational performance characterization in order to close the feedback loop between asset performance and all aspects of asset management, including plant design, predictive modeling and O&M.
Methods: The Single Model Characterization (SMC) framework consists of three components: data capability auditing, event management and operational characterization. Characterized events include outages, snow, curtailment, clipping and shade. Operational characterization utilizes inverter-level DC current and voltage data which enables isolation of key dependent to independent variable relationships. A key element of SMC is that it can be applied to both measured and predicted (e.g. PVSyst) data sets. This allows direct, quantitative comparison of all energy loss categories between real and virtual (simulated) assets. SMC provides the isolation of key performance drivers and effects required to enable the feedback loop for a robust continuous improvement process.
Results: SMC has been executed on over 2,000 (and growing) inverter-months of operational data. Results include an overview of both common and surprising data quality issues, detailed fleet benchmarking and application to the systematic improvement of predictive modeling accuracy.
Conclusion: SMC is a comprehensive characterization framework that has been developed over several years. We will share details of the characterization framework itself, results of performance benchmarking and application of the framework to numerous use cases.