Contingency-Based Resilience Assessment of Critical Utility Substations: An ETAP Framework for Accelerating Safe Interconnection of High-Density AI Data Center Loads

Abstract

This study examined the resilience of a critical utility substation under the proposed interconnection of a high-density AI data center load using a contingency-based ETAP framework. The study was motivated by the growing pressure placed on utility substations by concentrated digital infrastructure demand, particularly where safe interconnection depends not only on base-case capacity but also on the ability of the substation to remain within acceptable thermal, voltage, and protection limits during abnormal operating conditions. A quantitative nonexperimental case study design was adopted, and the analytical framework focused on measurable resilience indicators including transformer loading margin, bus voltage deviation, feeder loading utilization, short-circuit current magnitude, breaker duty margin, and contingency violation counts. The modeled system consisted of 2 power transformers, 8 buses, 6 outgoing feeders, 12 evaluated circuit breakers, and 10 credible contingency cases, with a final dataset of 12 retained operating states and 480 complete observations. The findings showed that the proposed AI data center interconnection materially reduced the resilience margin of the hosting substation. Mean transformer loading increased from 61.30% in the base-case condition to 79.80% in the post-interconnection normal condition and 86.70% under post-contingency operation, while the average reserve margin declined from 38.70% to 13.30%. Mean bus voltage deviation rose from 1.90% in the base case to 3.20% after interconnection and 4.80% under contingency, with the lowest recorded bus voltage falling from 0.981 p.u. to 0.941 p.u. Short-circuit current also increased from a mean of 21.70 kA to 24.10 kA and then to 28.10 kA across the same conditions. In parallel, breaker duty margin declined significantly, and contingency-related noncompliance rose from 0.00 violations in the base case to 1.00 in the post-interconnection normal state and 3.50 under post-contingency conditions. The results demonstrated that the substation remained broadly serviceable in selected normal states but experienced substantial reduction in disturbance tolerance after the large-load addition. The study concluded that contingency-based resilience assessment provided a rigorous basis for evaluating the safe interconnection readiness of critical utility substations exposed to high-density AI data center demand.