Headroom-Constrained Adaptive Power Oscillation Damping for Grid-Forming Battery Energy Storage Systems Under Modal-Estimation Uncertainty

Grid-forming battery energy storage systems (BESSs) can provide fast active-power modulation for oscillation damping, but adaptive action should be limited when modal evidence is weak or the damping-service power reserve is scarce. This paper develops an adaptive power oscillation damping (POD) supervisor that separates damping demand, modal-estimate reliability, and active-power headroom. Three bounded factors regulate only the adaptive gain increment upstream of actuator limits, leaving the baseline POD path and benchmark-specific estimator front end unchanged. Validation combines IEEE 9-bus factor ablations, transfer of the three-factor supervisory architecture and fixed threshold set to three New England 39-bus operating points, 100 paired physical-parameter realizations, and an independent average-value electromagnetic transient (EMT) benchmark. Across the paired realizations, active-power integral absolute error (IAE) is reduced by a median of 11.18% relative to conventional fixed-gain POD. Against a fixed controller matched to nominal supplementary-control effort within 0.05%, the median benefit is 0.272% (55 of 100 positive realizations; bootstrap 95% confidence interval, −0.037% to 0.720%); because this interval includes zero, a median performance advantage over the effort-matched comparator is not established. Estimator-attribution and rapid-derating tests show that early EMT action is prior-driven until online modal evidence becomes available and that the supplementary damping-command boundary is enforced by a final true-reserve clip independently of supervisory gain attenuation. For the seconds-scale service studied, active-power headroom can be more restrictive than stored-energy capacity.