Multicut Benders Decomposition for Generation Expansion with Dynamic Probabilistic Reserves

This study investigates the application of single cut and multicut Benders decompositions (BD) to the generation expansion problem, especially in renewable energy plants with time-varying dynamic probabilistic reserves (RPD). The research demonstrates that, even though RPD is represented as an exogenous and non-linear function, the problem of operation in the second stage remains convex. The RPD calculation considers the forecast error of renewable generation between consecutive hours, using a convex combination of the expected value and the Conditional Value at Risk (CVaR). Although the inclusion of RPD adds complexity, the convexity of the function is maintained, allowing for effective optimization through single cut and multicut DBs.