A Cycle-Based Framework for Quantifying the Operational Resilience of Pumped-Storage Hydropower from Reservoir Storage Dynamics

Pumped-storage hydropower (PSH) is widely evaluated using energy-based indicators that describe delivered output but provide limited information on how storage capability is depleted and restored during repeated operation. This study defines operational resilience for PSH as the capacity to sustain and restore generation capability over generation–recovery cycles. Normalised performance was derived from plant design tables linking upper-reservoir level to available generation energy, and a peak–trough–peak detector was applied to 17,453 hourly observations from a 1000 MW PSH plant in the Republic of Korea during 2019–2020. A total of 690 cycles were identified and characterised by robustness, rapidity and redundancy, with median values of 0.808, 0.155 and 0.800, values stable to within 0.01 across sixteen combinations of detection thresholds. Compensatory recovery, in which a deeper disturbance elicits a shorter recovery, is tested directly and is not observed: the association between disturbance depth and recovery duration is +0.103 (95% CI [+0.019, +0.200]) over all cycles and −0.053 ([−0.161, +0.063]) within the dominant overnight class, so recovery duration is either invariant to depth or weakly increasing in it. Recovery is instead schedule-driven. A schedule-matched operator blind to storage state produces a robustness distribution 2.7 times more dispersed than observed and does not reproduce the observed depth–duration association. The redundancy–rapidity relationship cannot be estimated in the benchmark because λ* carries no independent variation. Robustness and redundancy are structurally anti-correlated (r = −0.582, 95% CI [−0.642, −0.519]); an explicit two-reservoir mass balance confirms the mechanism, the regression of normalised lower reserve on normalised upper storage returning a slope of −1.005 against a closed-loop prediction of exactly −1, with the 11.8% residual accounting for catchment inflow and downstream release. The proposed resilience signature therefore complements conventional energy-based indicators by describing storage condition and recovery behaviour rather than generation volume or economic optimality.