Energy Digest

Quantum computers face limitations in solving power flow problems due to realistic grid properties, which cause the pseudo condition number of the DC susceptance matrix to grow polynomially or quadratically with network size. These limitations persist even when considering arbitrary bounded random line susceptances and various power flow scenarios, making end-to-end quantum advantage unlikely. Formal proofs are provided, including accompanying Lean 4 source code.

Bifurcation analysis reveals stability bounds and strong grid instability caused by sub-synchronous oscillations (SSOs) in grid-forming converter inner controllers, particularly with rapid onset of large oscillations past the Hopf bifurcation point. The study also investigates the impact of a circular current limiter, identifying spurious Hopf bifurcations in weak grids associated with SSOs under smooth approximations. This highlights the need for careful implementation of such approximations in grid-forming converters.

A new method for capacity accreditation that incorporates weather uncertainty into variable renewable energy forecasts provides more accurate and reliable capacity credits, yielding better investment signals than traditional deterministic or averaged approaches. This method reduces distortion in resource expansion decisions and improves reliability outcomes. Inertial methods can lead to material reliability worsen outcomes compared to incorporating weather information.

Traditional imbalance-management methods for supervised learning in predictive maintenance are severely limited due to their assumption of homogeneous minority populations. A new approach using a specialized multi-generator GAN architecture improves the identification of infrequent failures by producing more realistic minority samples. This method outperforms traditional resampling methods and individual-generator GAN augmentation, achieving higher PR-AUC and recall scores.

A dataset of 15-minute smart meter measurements from 2,447 residential installations in Switzerland provides real-world data on evolving electricity consumption patterns driven by renewable energy adoption and electric vehicles. The dataset includes active and reactive energy data for diverse installation types and a subset of customers participated in a pilot project with novel grid tariffs and automated load control. The dataset aims to support research on distribution grid modeling, load forecasting, and the assessment of time-varying tariffs.

The article develops a large-signal stability analysis for a sampled-data optimization-based secondary controller for distributed energy resources, providing computable bounds on voltage, filtered reactive power, and the secondary control input. The analysis establishes how optimizer objectives and constraints connect voltage regulation with equal per-unitized reactive power sharing and investigates input-to-state stability of frequency dynamics. It provides a rigorous mathematical foundation for sampled-data optimization-based secondary control of distributed energy resources.

Agentic artificial intelligence systems are increasingly proposed for critical engineering domains where decisions carry significant consequences. A new study proposes a trustworthiness model with five dimensions (safety, robustness, transparency, accountability, and privacy) and identifies recurring design patterns, shared failure modes, and domain-specific gaps across four constraint-bound engineering domains. The research outlines a reusable, cross-domain assurance framework to address agentic AI trustworthiness in critical systems.