Optimizing Virtual Power Plants with Parallel Simulated Annealing on High-Performance Computing

This work focuses on optimizing the scheduling of Virtual Power Plant (VPP) to maximize social welfare and improve energy trading with the grid. Considering distributed energy resources (DERs) and battery lifespan limitations, we address the problem as a Mixed Integer Linear Programming (MILP) using a parallelized Simulated Annealing algorithm implemented on High-Performance Computing (HPC). This parallelization accelerates the exploration of the solution space, improving computational efficiency and enabling the system to handle larger networks of DERs and more complex scheduling decisions. The approach is evaluated through rigorous simulations, highlighting its effectiveness in reducing optimization time while maintaining high-quality solutions. The results demonstrate that our method significantly enhances the scalability of VPP scheduling and facilitates more responsive and efficient energy market participation, particularly in scenarios with stringent operational constraints and dynamic energy pricing. All experiments were conducted following a systematic pipeline, including hyperparameter optimization, to ensure that the configurations were optimized for efficient and effective performance.