Estimation and forecasting methods for design and operation of photovoltaic plants stars

Resum

The growing awareness of the tremendous environmental impact of burning fossil fuels is promoting the shift to renewable energies sources to generate energy. Solar energy, along with wind energy, is forecasted to become one of the main energy sources in the energy mix of the future. Among all the technologies available to transform solar irradiation to electricity, solar photovoltaic (PV) stands out as the most developed and promising, due to its simplicity and relative ease of maintenance. It has witnessed an incredible reduction in prices and its efficiency in commercial applications has rocketed. Consequently, solar PV has emerged as a leading technology in the transition to a more sustainable future for many countries. Because it has not reached maturity yet, there are still numerous areas of research concerning the development of PV. This thesis deepens the study and development of PV technology throughout its life cycle, focusing on the design and operational stages. The thesis includes four different studies, three of them tackling one major issue within each stage and one review of the literature. With respect to the design and planning stage, the estimation of solar irradiation is of great concern, since it is the main input to a PV plant. Accurate estimates of the solar resource leads to increased revenue and a reduction of uncertainty during operation. Given that ground measurements of solar irradiation are scarce, we have developed a methodology using machine learning techniques to estimate solar irradiation from other more commonly measured meteorological variables and then geostatistical techniques were applied to obtain maps of continuous annual irradiation values. Regarding the operational stage of a PV plant, we have focused on two aspects: how to increase the value of the electricity and how to increase production. The former issue was addressed through forecasting of electricity production from a PV plant. Improved forecasting leads to increased revenues. We studied the value of forecasting in the electricity market and the margin for improve ment. Tracking strategies were utilized to investigate how production could be increased. During cloudy and overcast weather conditions most irradiation comes from its diffuse component. Because of this, we analyzed the potential for irradiation increase derived from a tracking strategy that sets PV panels facing the zenith when those conditions are present. Additionally, an operational algorithm was developed to benefit from the tracking strategy proposed. In summary, this thesis helps advance the collective knowledge surrounding solar technology in an effort to guide the transition towards a more sustainable future.