Uri Precursor Window and Operator Decision Rule: Quantitative Characterization from Real ERCOT Data and Translation into Actionable Guidance for Balance Authorities

This paper presents a quantitative characterization of the structural precursor window preceding the ERCOT Winter Storm Uri load-shed onset (15 February 2021, 01:21), derived entirely from ordinary metered data — native hourly system load and Fuel Mix Report generation, with no reconstructed or synthetic series used. Two independent observables, signed imbalance mean and demand autocorrelation, are shown to converge on a dual-horizon precursor structure: a first detectable excursion at approximately 5.2 days before rupture, a zero-crossing structural lock-in at 3.47 days, and a final irreversible intensification at 3.4 hours before rupture. The signals are shown to survive several real-world reporting distortions — unit inconsistency, telemetry degradation, diurnal contamination, and window leakage — that are typically absent from synthetic grid testbeds. The paper translates these measured lead times into a simplified two-trigger operator decision rule, mapped directly onto demand-response products ERCOT already operates (ERS, RRS, ECRS), requiring no new market design. The choice of demand autocorrelation as the primary structural observable is justified against several alternative candidates (variance, spectral slope, Hurst exponent, DFA, entropy, AR-coefficient tracking), with citations to the relevant critical-transitions and power-systems literature. Decision thresholds (24-hour confirmation window, −1σ intensification boundary) are explicitly identified as engineering choices calibrated to the observed structure of this event, not as optimized or validated parameters. The work is scoped as a single-event, single-topology characterization. The paper states plainly what remains untested: Trigger 1’s false-positive rate against ordinary non-crisis winters, threshold sensitivity, and generalization beyond thermally-stacked grid topologies like ERCOT. Independent, non-overlapping evidence from a separate national outage-record analysis is cited narrowly as consistent with — but not confirmation of — the generality of the reported structure.