Progress in regional PV power forecasting: A sensitivity analysis on the Italian case study. (April 2022)

• Record Type: Journal Article
• Authors: Pierro, Marco; Gentili, Damiano; Liolli, Fabio Romano; Cornaro, Cristina; Moser, David; Betti, Alessandro; Moschella, Michela; Collino, Elena; Ronzio, Dario; van der Meer, Dennis
• Is Part Of: Renewable energy. Volume 189(2022)
• Page Start: 983
• Page End: 996
• Publication Date: 2022-04
• Subjects: PV plants -- Regional PV power Forecast -- Upscaling forecast methods; Renewable energy sources -- Periodicals; Power resources -- Periodicals; Énergies renouvelables -- Périodiques; Ressources énergétiques -- Périodiques; 333.794
• DOI: 10.1016/j.renene.2022.03.041
• Languages: English
• ISSNs: 0960-1481

Abstract

Abstract: The increasing penetration of PV generation, driven by climate strategies and objectives, calls for accurate production forecasting to mitigate the negative effects associated with inherent variability, such as overgeneration, grid instability, supplementary reserve request. The regional PV power forecasting is crucial for Transmission and Distribution system operators for a better management of energy flows. In this work many aspects of regional PV power forecasting are investigated, by means of a comparison of six different forecasting models applied to predict the hourly production of the following days on six Italian bidding zones for one year. In particular, the work shows that the forecasting accuracy is mainly affected by the algorithm and its pre and post processing, with a range of 30% in performance accuracy, while it is less impacted by the forecasting horizon. It has been verified that the accuracy in the irradiation prediction, used in input to the power forecasting algorithm, has less impact compared to single plants. The work confirms the performance improvement which can be obtained by increasing the size of the area to which the prediction refers, through a comparison between the forecasting at bidding zone and national level. Finally, we show that the larger the controlled forecast area, the smaller the impact on the forecast accuracy due to the non-uniform spatial and capacity distribution of the PV fleet. This means that as the size of the region increases, the average irradiance progressively becomes the best PV power predictor. We refer to this phenomenon as: "input smoothing effect".