Stochastic Multi-Fidelity Scheduling of Flexibility Reserve for Energy Storage

Abstract

This paper proposes a continuous-time two-stage stochastic optimization model for multi-fidelity co-optimization of energy and flexibility reserve provided by generating units and energy storage (ES) devices in day-ahead operation. The flexibility reserve, defined as a single continuous-time trajectory that combines the balancing and ramping reserves, not only supplies the energy deviation but also the ramping requirements of load and renewable generation in power systems operation. The proposed model co-optimizes decision variables with different modeling fidelity, where the energy and flexibility reserve schedules are modeled and optimized by Bernstein polynomials of different degrees to match the flexibility requirements of load and renewable generation in day-ahead and real-time operation stages. Numerical studies, conducted on the IEEE reliability test system with the load and solar data of California ISO, highlight the benefits of the proposed stochastic multi-fidelity model over traditional discrete-time models in efficient utilization of ES flexibility to supply the energy and ramping requirements of the net-load and avoid scarcity events.