Methods to incorporate correlation of intermittent renewable generation and load into non-sequential Monte Carlo simulation for reliability evaluation

This paper compares four methods to incorporate the existing correlation between intermittent renewable generation (such as wind, solar, hydro, etc.) into Monte Carlo simulation (MCS) using the state sampling approach (also named as non-sequential MCS) for power system reliability studies. The methods investigated are employed for non-parametric distributions and comprise those based on the correlation matrix (Iman-Conover rank correlation method and Nataf Transformation) and on the copula theory for multivariate cases (Vine Copula and Hierarchical Archimedean Copulae). Four case studies are analyzed (two from the literature and two from the Brazilian system), and the methods are compared using statistical metrics to evaluate their ability to accurately reproduce the original multivariate distribution. Afterwards, they are compared when applied to reliability evaluation by non-sequential MCS, both in terms of accuracy by comparison with sequential MCS indices and of computational burden required by each one. The comparisons allow identifying which methods are most suitable for the type of study and data considered.