Solar photovoltaic energy scenarios generation: a novel methodology for multi-area electricity markets

Abstract

In this paper, a novel methodology is introduced for the generation of solar photovoltaic energy scenarios within multi-area electricity markets. Diverging from the existing short-term focus in the literature, the proposed approach tackles the challenge of creating realistic long-term scenarios for solar energy, taking into account the inherent variability of solar availability and the correlation between different areas. A distinctive feature of this methodology is the segmentation of historical data on a monthly basis and the application of unique Seasonal Auto Regressive Integrated Moving Average (SARIMA) models for each month and area. The accuracy of scenario generation is further enhanced by the production of hourly multivariate residuals. These residuals, being correlated and conforming to normal distribution functions, integrate spatial correlations derived from covariance matrices of monthly historical data series. Then, Monte Carlo simulations are employed, effectively merging the temporal dependencies captured by the SARIMA models with the spatial dependencies gleaned from historical data covariances. This integrated process results in more comprehensive and reliable solar energy scenarios. The focus of the study, encompassing Spain, Portugal, and France, involves an evaluation of the scenario quality over a two-year period. Main results indicate that the inclusion of a monthly-level dependencies modeling significantly enhances the quality of solar power scenarios, thereby improving their applicability across various temporal scales.