Data-driven scheduling of a grid-connected university campus battery energy storage system considering variable weather and energy pricing

BESS can provide many crucial grid services including voltage and frequency support, however, capital costs can hinder widespread installation. Thus this work develops a data-driven MILP approach for BESS dispatch to enhance consumer cost recovery through energy arbitrage and mitigate the barrier of high capital costs. The proposed algorithm uses location-relevant weather data, to forecast PV generation and load consumption, as well as market data to create a BESS dispatch schedule. The proposed BESS algorithm is validated on a Raspberry Pi 4b microcontroller using real-time HIL simulation for 72 h on a real-world-based university campus microgrid consisting of a 9 MWh/3 MW BESS and 1.4 MW DC/1.1 MW AC PV solar canopy. Economic analyses indicate the proposed algorithms ability to decrease system energy costs when using LMP pricing, translating in a 7.27% decrease in LCOE compared to a 3.12% to 3.20% decrease when using an RB scheduling method. Similarly, LCOE is reduced by 5.32% when using TOU energy pricing. Additionally, a sensitively analysis of PV and load forecasting error with error biases ranging from −10% to 10% indicate a resulting increase in LCOE of 0% to 0.674%. These key findings indicate increased economic viability when implementing the proposed dispatch approach. • Development of a MILP-based data-driven BESS scheduling control for use in a grid-tied microgrid. • Validation through a 72-hour continuous real-time C-HIL simulation of a real-world microgrid. • Performed economic comparison of electricity pricing yielding 5.32%–7.27% reductions in LCOE. • Sensitivity analysis of forecast error demonstrates reduction in performance by 0%-0.674%.