Two-step optimised energy model for the performance evaluation of two agrivoltaic plants

Nowadays, PhotoVoltaics (PV) has become one of the fastest-growing renewable energy technologies and plays a key role in the ongoing energy transition. Conventional PV systems rely on energy conversion from the front side of the module only, which limits the achievable efficiency. For this reason bifacial PV technology has emerged in the last years as it allows the simultaneous collection of irradiance on both front and rear sides of the modules, hence increasing the energy yield and improving the competitiveness of solar energy. In particular, a promising solution for the use of bifacial modules is agrivoltaic technology, which integrates solar energy production with agricultural activities. Despite these advantages, bifacial systems introduce additional challenges, particularly in modelling, monitoring and assessing performance under real operating conditions. Therefore accurate tools are needed to estimate key parameters and to detect possible sources of underperformance which compromise the plant operation. In this thesis work an optimised energy model has been determined for the underperformance evaluation of two multi-MW agrivoltaic plants with bifacial modules and east-west sun tracking systems. The plants are both located in Sicily, respectively in Mazara del Vallo and Paternò (Italy), and are managed by the Italian company ENGIE ENERGIES ITALIA S.r.l.. The analysis has been performed on April 2023 data for the Mazara del Vallo PV plant and on July 2023 data for the Paternò plant. The procedure, implemented in MATLAB ambient, performs a two-step optimisation of critical coefficients. In the first step, the current efficiency and particularly the bifaciality factor (BF) are estimated, while in the second step the global DC/AC efficiency and the thermal coefficient γ are evaluated exploiting the results derived in the first stage. The process is carried out at stringbox level, after applying a dedicated set of filters to the initial dataset in order to exclude measurements affected by low irradiance, clipping effects, or unstable weather conditions. The effectiveness and robustness of the optimisation process are assessed through the calculation of the Normalized Root Mean Square Error (NRMSE). The work has required the development of a Graphic User Interface (GUI) again in MATLAB ambient, this time using a dedicated software called App Designer. This tool has been useful to find any source of underperformance inside each plant by evaluating deviations with respect to the model estimations at stringbox level. The results are visualised thanks to specific heatmaps which visually show the presence of any faulty stringbox, thus contributing to system underperformance and that may require intervention.