A Two-Stage Algorithm for Optimal Scheduling of Battery Energy Storage Systems for Peak-Shaving

Increased penetration of Renewable Energy Sources (RES) with intermittent and variable power output has led to increased use of Battery Energy Storage Systems (BESS) for grid applications. This paper presents a two-stage algorithm for optimal energy scheduling of BESS interfaced with RES. Initially, a multivariate linear regression-based estimation of voltages, currents, and network active power loss is performed using a synthetic dataset generated from the network model. Thereafter, a linear programming (LP) formulation is used to determine the output power of the BESS aimed at maximum peak-shaving and valley-filling, based on predicted day-ahead net demand and solar photovoltaic (PV) output. BESS technical and experimental constraints are considered in the model for an improved lifetime of the batteries. Compared to nonlinear approaches, the linearized model would reduce computational complexity and time, while maintaining reasonable accuracy. The linear programming model is solved using MATLAB, and the proposed algorithm is implemented on a real-world distribution feeder modeled in OpenDSS. The results show a significant reduction in peak demand, net demand variation range, and voltage variability caused by intermittent PV output.