SCARO: Scenario-Conditioned Risk Optimization for Day-Ahead Battery Energy Storage Scheduling

Abstract

Day-ahead battery arbitrage requires assessing the return and downside risk of complete charge–discharge cycles. We propose Scenario-Conditioned Risk Optimization (SCARO), which corrects supply and demand forecasts and maps jointly resampled whole-day residuals to price paths. An exact planner identifies adverse outcomes separately for each feasible schedule and combines mean and lower-tail profit to select an operating cycle or no operation. A 182-day rolling evaluation in the Western Inner Mongolia real-time electricity market compares SCARO with 21 baselines for a battery with zero initial and terminal states of charge and at most one daily cycle. The configurations use a single price model (SCARO-S) or ensembles (SCARO-E and SCARO-L), with the ensemble configurations differing in boundary correction. SCARO-S achieves the highest mean profit in 13 of 16 efficiency–cost conditions. At nominal efficiency and zero cycle cost, its mean and worst-month profits exceed the strongest baseline for each metric by 4.72% and 8.76%, respectively; SCARO-L achieves the highest oracle capture. With forecasts held fixed, lower-tail scheduling improves mean profit and loss frequency for SCARO-E and SCARO-L relative to their risk-neutral controls.