Energy Digest

Dynamic operating envelopes (DOEs) have been developed to publish time-varying export limits that keep distribution networks within operational limits. A new two-stage, multi-period framework has been proposed that addresses fairness by redistributing capacity through cumulative proportional fairness and capping maximum curtailment ratios, resulting in fair DOEs with minimal impact on curtailment. The approach achieves high Jain fairness indices close to unity while maintaining AC voltage deviations below 0.01 p.u.

This paper proposes a framework to analyze the stability of plug-and-play distributed energy resource systems under delay, revealing a fundamental trade-off between theoretical stability limit and practical hosting capacity. The analysis shows that the admissible range of connectable units decreases with increasing normalized delay, leading to a non-monotonic dependence on delay. A feasibility boundary may emerge beyond which no admissible system size exists due to implementation constraints.

Unit-to-Plant Stability Shaping of Multi-Electrolyzer ReP2H Plants via Interface Design and Dispatch: A stability-oriented framework is proposed for controlling Alkaline water electrolysis units to prevent oscillations caused by coupling between rectifier control and electrolyzer dynamics. Higher loading reduces stability, while larger dc-link capacitance and higher Buck bandwidth improve it. The plant-level model distinguishes the stability margins of different power allocations.

The proposed probabilistic active learning framework successfully expands the transient stability region of attraction for networked grid-interactive inverters by systematically driving the estimated stability boundary outward through intelligent coupling of electromagnetic transient simulations and uncertainty-guided frontier search. This approach substantially reduces estimation conservatism, achieving a $20$-fold volumetric enlargement of the certified stability region compared to classical baselines. The methodology requires at most 220 time-domain simulation queries per system.

An open phase detection method using synchronized phasors to calculate connection impedance between two measuring points has been developed for distribution systems. The method detects open phases by analyzing the positive-sequence impedance, which becomes significantly larger and negative under internal open phase conditions. This new method is scalable, cost-effective, and reliable with minimal impact from back-feed.

A physics-guided generative design framework was developed for optimizing liquid-cooling channels in high-power multi-chip packages, reducing maximum GPU temperature by 33.6%, GPU temperature spread by 52.5%, and pressure drop by 72.8% compared to a conventional reference topology. The framework generated over 2,200 designs before identifying the thermally leading feasible design, G1016, which showed promising results in both reduced-order thermal-hydraulic models and three-dimensional conjugate heat-transfer simulations. These findings demonstrate the efficiency of physics-guided generative design in discovering innovative cooling channel architectures for advanced packaging applications.

An edge-based detection method is developed for inferring correlated AI data-center load episodes from electrical measurements alone, without access to data-center telemetry. The method analyzes cross-facility power measurements and uses pairwise power correlations as the discriminative signal, achieving real-time execution on commodity edge hardware. This approach can help grid operators detect spatially correlated load variations in sub-second power swings and separate independent and correlated cases.