High-renewables grids are planned in min but judged in milliseconds; credible studies must therefore resolve both horizons within a single model. Current adequacy tools bypass fast frequency dynamics, while detailed simulators lack multi-hour optimization, leaving investors without a unified basis for sizing storage, shifting demand, or upgrading transfers. We present a two-layer Hierarchical Model Predictive Control framework that links 15-min scheduling with 1-s corrective action and apply it to Germany’s four TSO zones under a stringent zero-exchange stress test derived from the NEP 2045 baseline. Batteries, vehicle-to-grid, pumped hydro and power-to-gas technologies are captured through aggregators; a decentralized optimizer pre-positions them, while a fast layer refines setpoints as forecasts drift; all are subject to inter-zonal transfer limits. Year-long simulations hold frequency within ±2 mHz for 99.9% of hours and below ±10 mHz during the worst multi-day renewable lull. Batteries absorb sub-second transients, electrolyzers smooth surpluses, and hydrogen turbines bridge week-long deficits — none of which violate transfer constraints. Because the algebraic core is modular, analysts can insert new asset classes or policy rules with minimal code change, enabling policy-relevant scenario studies from storage mandates to capacity-upgrade plans. The work elevates predictive control from plant-scale demonstrations to system-level planning practice. It unifies adequacy sizing and dynamic-performance evaluation in a single optimization loop, delivering an open, scalable blueprint for high-renewables assessments. The framework is readily portable to other interconnected grids, supporting analyses of storage obligations, hydrogen roll-outs and islanding strategies.