Accuracy validation of a model, based on weather forecasts, for the hourly power calculation of photovoltaic systems

Global warming and the resulting climate change are some of the biggest challenges and problems of today. In this context, a central role is played by energy transition, that is the transition from energy sources that release larger amounts of carbon dioxide, such as fossil fuels, towards clean energy sources. In this perspective, the contribution of photovoltaic energy is fundamental, in particular in Italy, where its abundance is relatively high compared to European average. In addition, renewable energy sources, being inherently decentralised, can contribute to energy independence, which is a very important topic after the huge increase in gas and electricity prices in Europe last year and recent geopolitical issues. On the other hand, renewable sources also have disadvantages; among these the main one is undoubtedly the intermittency of the resource and the consequent difficulty to forecast the production profile, which is determined by external events, while the production output of traditional plants is simply determined by energy demand. Electrical storage can be a solution as a compensation of the variability of the energy resource, however, it would require higher costs and materials, whereby it is not feasible on a very large scale. Forecasting photovoltaic production, which is the topic of this thesis, is a complementary alternative to storage and is essential to integrate renewables into the electricity grid and minimize costs. The aim of this thesis is to utilize a photovoltaic power calculation model whose result is on an hourly production profile, with the use of weather forecast variables as input data to predict the produced electrical power. The quality of the model is tested on a PV plant for which electrical power and total irradiance are measured. The above-mentioned models are optimized in a second step to reduce the final error. As forecasts nature incorporates high uncertainties, a main focus of this thesis is represented by decoupling weather forecast error from power calculation model error, and by evaluating forecasts quality. More in detail: -??In Chapter 1 an overview of photovoltaics is presented, which describes the technology used, its contribution to the electricity grid, the importance that has recently acquired in the context of global warming. In particular the equation of PV cells which correlates current, voltage and power with weather parameters like irradiance and air temperature. Finally, a review of the literature of some photovoltaic prediction models., with the focus on the difference between physical models, that use the correlation between meteorological and electrical variables, and statistical models which are based on statistical methods and artificial intelligence. -??The second chapter illustrates what will be the calculation model used in this thesis, which belongs to the category of physical models, and how it will be optimized in a second step. -??Chapter 3 is dedicated to the description of the methodology used to acquire data. In detail, in the study there will be the contribution of measured irradiation data, electrical power data measured and expected meteorological data (in particular irradiance components, but also temperature and wind speed). The greatest attention is paid to this last type of data, since they are forecast data, they will have to be continuously updated and they introduce the highest source of uncertainty. Some daily irradiation profiles will be analyzed in this section. -??Chapter 4 will report the results