Two-Stage Stochastic Optimization of Multi-Service Behind-the-Meter BESS Sizing and Scheduling

This paper presents a risk-aware co-optimization framework for the sizing and day-ahead scheduling of behind-the-meter battery energy storage systems (BESSs) under renewable and load uncertainties. For each day of a year, photovoltaic and load scenarios are generated using quantile-based forecasting, followed by Monte Carlo sampling and clustering to preserve probabilistic diversity with tractable complexity. The representative daily scenarios generated over one year are aggregated into an annual scenario set and then used in a two-stage stochastic optimization model. The first stage determines the BESS power and energy capacities, while the second stage optimizes hourly dispatch under each scenario. A Conditional Value-at-Risk (CVaR) term is incorporated into the objective to capture rare but high-cost events, balancing expected cost and tail-risk exposure. Compared to the no-storage baseline, the optimized 16.6 kW / 33.1 kWh BESS reduces total annual cost by €4.6k (5%) while ensuring robust operation across all scenarios. The study provides a computable analytical foundation for achieving a balanced trade-off between economic performance and operational robustness in BESS design under renewable and load forecast uncertainties.