Análise de fontes de incerteza no levantamento e estimativa da capacidade de geração solar fotovoltaica

Abstract

This dissertation investigates the sources of uncertainty in estimating the photovoltaic (PV) solar energy generation capacity in the state of Paraíba, considering georeferenced data, grid failures causing events in PV inverters, and variations in climatic variables. The research adopts a quantitative approach, focusing on the reliability of different meteorological data series to understand their impact on PV generation predictability. A comparison was conducted among geoclimatic databases, considering data from sources such as INMET, NASA-POWER, and OPEN-METEO, contrasted with real measurements from installed systems. The results showed that INMET and NASA-POWER tend to have a strong correlation with real measured data, while OPEN-METEO exhibited a lower correlation. This finding highlights the importance of selecting meteorological databases that accurately reflect reality to improve PV generation capacity estimates. Additionally, an empirical study was conducted to assess the impact of grid failures and their effects on PV inverters' efficiency. Real operational data from the studied inverters were analyzed, revealing that months with failure occurrences experienced a generation capacity efficiency loss between 10.03% and 14.67%, whereas months without failures had losses of only 2.55%. Based on these findings, an equation was proposed that incorporates inverter downtime due to grid failures into PV generation estimates, enabling more accurate forecasts that benefit system designers and integrators. The impact of climatic variables on PV generation was analyzed over a decade, considering temporal variability. The analysis indicated that increasing temperature and decreasing wind speed negatively affect PV generation. The statistical model demonstrated that 65% of the variability in PV generation capacity could be explained by these fluctuations, emphasizing the importance of incorporating them more rigorously in technical and economic feasibility studies. Thus, this dissertation contributes to understanding uncertainties in PV generation forecasting, providing insights to optimize the selection of meteorological databases, mitigate the impacts of inverter downtime due to grid failures, and refine forecasting methodologies, ultimately supporting more efficient PV system planning.