Adequacy, flexibility and other technical-economic indices used to support planning choices in the electrical system over long-term horizons are strongly influenced by Non-Programmable Renewable Sources (NPRS), as well as by other stochastic quantities such as load. It is therefore necessary to accurately represent their characteristics in order to evaluate these indices. This requires tools capable of generating several plausible time series of stochastic quantities that feed the Monte Carlo cycles of the electrical system analysis tools. Space-time dependencies between random quantities and rare events are of special importance in this context. Furthermore, since long-term time horizons are examined, it is necessary to take into account climate changes, due to their potential impact on some of the involved quantities.

Following an analysis of the literature and available data repositories, a methodological approach was developed to generate plausible time series of wind, PV and load power generation, which takes into account the above-mentioned aspects.

After specific pre-processing for each type of data, the stochastic part of the time series is identified through a p-order Vector Autoregressive model VAR(p). The latter then generates a number of time series which, after post-processing, have the same statistical characteristics as the input series. The simulated series preserve individual aspects (distribution, mean, variance, dependence on the past), but also cross-correlations and the frequency with which rare events occur. The pre-processing of the wind power series consists in dividing them by season and creating a separate model for each one of them. In photovoltaic power series, seasonality is extracted by considering the mutual position of the Earth and the Sun. The load seasonal component is represented by three profiles depending on the type of day, and the random part is influenced by temperature, which is particularly affected by climate change. The initial implementation of the tool for the wind power case demonstrates how the simulated series preserve the statistical properties of time series.