Simulation with performance evaluation of photovoltaic production using long term monthly irradiation data

To adequately calculate photovoltaic (PV) energy production in most applications, long term hourly or sub-hourly solar radiation data are required. When such data is expensive or not easy to acquire, treat or handle, synthesis models could be employed with long term monthly data. However, there is no conclusive work about the effects on the quality of the result when a chain of synthesis, decomposition, transposition, and other models is used to estimate energy production using monthly input data. This article proposes a methodology that allows to select ensembles with well-integrated models for various applications. Using these techniques, an overview of the performance of seventy decomposition, transposition and synthesis models is presented. Besides, good model chains are selected, and the results validated through a comprehensive simulation procedure conducted for several climatic types. Tens of thousands of combinations of synthesis, decomposition and transposition models are simulated and compared to reference values. It concludes that, for some applications, correct treatment of monthly average data can bring results that are as suited for the purposes as multi-annual hourly data, but with the advantage of using more accessible and manageable data, in some cases coming from more accurate databases.