Model Predictive Optimal Dispatch of Behind-the-Meter Energy Storage Considering Onsite Generation Uncertainties

This paper presents the model-predictive optimal dispatch of a behind-the-meter energy storage (BMES) system considering onsite generation/load variabilities and forecasting uncertainties. First, onsite generation and consumption are forecasted for a given facility with different confidence level using auto-regressive integrated moving average model. Subsequently, the cost-optimal dispatch of BMES is computed considering the forecasting uncertainties, cost of energy, cost of battery degradation, and behind-the-meter (BTM) services. In particular, the BMES is deployed for multiple BTM services, including peak-load reductions, smoothing intermittencies from onsite renewables, and load shaping of given facility. A mixed-integer non-linear programming based optimization is formulated and solved in GAMS using KNITRO solver to compute the cost-optimal BMES dispatch. The performance of the proposed method is investigated through a 24-hour time-series simulation in a co-simulation environment (GAMS, MATLAB, and R) using operational data of a residential consumer. The results demonstrate that the proposed method can simultaneously maximize BMES operational benefits and provide insights for sizing resources to compensate power imbalances of a facility.