Photovoltaic solar energy conversion: prediction models applied to tracking and fixed-tilt systems at European scale

The present Ph.D. thesis aims to present the development of models for diagnostic of large photovoltaic plants, the maximization of the solar radiation captured by ground-mounted bifacial plants, both fixed and single-axis tracking, and the optimization of the exploitation of the solar resource in urban-integrated systems. The first part of the work concerns the development of robust models for predicting the performance and enabling automated monitoring of large-scale photovoltaic systems, utilizing historical and real-time electrical and thermal data. With the growing global presence of large PV installations, the ability to detect and forecast failures or malfunctions is increasingly critical for timely maintenance decisions. The proposed model simulates the behavior of PV modules and inverters under varying irradiance and temperature conditions and can predict shading effects that impact energy output. Validated with real electrical data from six large PV plants in southern Italy, the model can predict real-time power generation with an error margin of 4.1%. More importantly, the model has proven highly effective in identifying downtime due to issues with inverters or string failures through sub-hourly comparisons over several years. Simulations and field measurements indicated that energy losses due to grid coupling downtime and shading can be accurately identified and differentiated from component malfunctions, reducing false alarms. Additionally, aerial infrared imaging further verified the model's capability in detecting failures, understanding the link between thermal anomalies and system underperformance, and predicting the annual degradation rate of PV plants. The second topic of this thesis regards the development of a Matlab model for the detailed calculation of the irradiance distribution on ground and on both sides of bifacial PV modules arrays considering the irradiance distribution of the sky vault, the presence of multiple PV rows and the reflection from the ground. The 3D celestial vault is represented using the Perez "All-Weather" sky model, accounting for different types of radiance distribution across the sky dome. The model uses view factor calculations for each sky element, as well as for each different subareas of the modules and ground. Results have been validated through comparisons with simulations using NREL’s "Bifacial_radiance" tool. The detailed calculation of irradiance on the front and rear surfaces of the modules provides accurate estimates of available solar energy and supports strategies to maximize output, considering cell-to-cell mismatch and the effects of the "bifaciality factor" in adjacent arrays. This calculation tool has been used to determine the best backtracking tilt angle for single axis tracking bifacial plants. It has been done with the approach of maximizing the incident radiation on both front and rear side of modules during the operation of the system, considering the plant’s geometry and instantaneous position of the sun and weather condition. An hourly, year-round analysis was performed for both Mediterranean and continental sites to compare the solar yield from this model’s tracking algorithm against a backtracking strategy based on sun position and different azimuthal orientations. The model shows that solar energy collection can increase by up to 6% during months with higher diffuse irradiance. Another aspect of this thesis is related to the creation of a European map of the best tilt of fixed south-facing bifacial PV plants on regular 50x50km grid (2382 locations). For each point, TMY solar data are retrieved from the PVGIS platform and a 3D yearly cumulative distribution of radiance on sky has been built. Thanks to the aforementioned model for radiance distribution on modules an iterative optimization process has been used to determine the optimal tilt angle that maximizes energy yield from both the front and rear surfaces of the photovoltaic modules. This optimization considers direct, diffuse, and reflected irradiance, with ground albedo and the spacing between PV rows included as key parameters. Additionally, a free web-based tool has been developed to enable users to calculate the optimal orientation for photovoltaic systems at any longitude and latitude within the European Union.