Toward Efficient Robust Dynamic Dispatch: A Two- Stage Model With Tight Upper Bound Guarantees

Canonical two-stage robust dispatch models ignore non-anticipativity, resulting in radical here-and-now decisions and a lower cost bound. In contrast, non-anticipative variants impose excessive constraints, yielding an upper bound and conservative scheduling. This letter proposes a two-stage model that narrows the gap between these bounds relative to the exact multistage formulation. The core idea is to construct enlarged robust operating regions for dispatchable resources by systematically modeling prior decisions, such as generator output and state of energy, as decision-dependent uncertainties. This unified approach better balances model fidelity and computational tractability in robust dynamic dispatch.