Evaluating the performance of a photovoltaic (PV) system under varying, real-world environmental conditions is important to many stakeholders across residential and commercial sectors and utilities. System owners want to know that they are capturing as much energy as possible. Operations and maintenance contractors need to know where and when to perform corrective maintenance. For large scale commercial and utility systems, sale contracts increasingly include some form of a power guarantee, in which the seller pays the buyer for system underperformance after installation. However, detailed and accurate performance analysis is not available for the smaller residential and commercial systems.
The toolbox of performance analysis for utility PV power plants include capacity tests (Whaley 2016), short-term performance ratio tests, performance index tests (Dierauf 2013), or long-term energy tests (Kurtz 2013). Additionally, industry performance engineers commonly perform more detailed retrospective analysis of system loss factors. Each of these tests requires that system data be manually collected or verified, and often rely on expert opinions. It is rare to see these analyses performed for other asset classes due to the time/cost of performing the work, issues with data quality and quantity, and the domain knowledge necessary to accurately carry out the work. Even when they are applied under the best conditions, these approaches have the following deficiencies:
1.Rely on deterministic models based on local irradiance measurements that do not account for uncertainty in model parameters or input data.
2.Do not extract trends in observed data that quantify and explain low system performance.
3. Lack standard methods for calculating system performance metrics, in particular loss factor, and capacity factor.
The goal for PVInsight is to develop an open-source toolkit written in Python, which automates loss factor estimations for small and medium-sized systems, while addressing the issues listed above.