Energy Digest

Prosumers can cooperate voluntarily under strategic self-interest to achieve effective demand response in smart microgrids through the use of Large Language Model (LLM) mediation. The LLM-compiled directives generate realistic prosumer behavior with baseline cooperation near 50% and clear differentiation under influence, outperforming unstructured messaging and a no-intervention baseline. Structured LLM compilation, grounded agent reasoning, and network-aware targeting are complementary design principles for scalable, interpretable demand-response coordination in smart-city energy systems.

Multi-terminal DC transmission systems based on modular multilevel converters (MMCs) rely on safely regulating MMC currents under various operating conditions to ensure reliable operation. A new design framework, leveraging Lyapunov stability conditions, proposes a less conservative static state-feedback controller that addresses the challenge of achieving fast transient response while maintaining robustness against uncertainties. The proposed method is assessed and compared with existing methods using the CIGRE MT-HVDC benchmark simulation results.

Offshore energy hubs use grid-forming modular multilevel converters to enable large-scale offshore wind integration. A unified current-limiting strategy combining variable virtual impedance and virtual-power mechanisms improves transient stability by limiting fault currents and maintaining synchronism during severe faults. The proposed strategy also enables automatic power redistribution among converters, improving the overall stability margin of fully converter-based OEHs.

A novel dynamic optimization algorithm is proposed to optimize virtual inertia and damping in converter-based power systems, addressing stability, cost-efficiency, and resilience concerns. The algorithm considers performance metrics and disturbance magnitudes to achieve optimal allocation of these parameters. It is validated on a three-area system and compared to an existing $\mathcal{H}_2$ system-norm-based approach.

Temperature-aware heat pumps are expected to significantly increase peak electricity demand, but building thermal inertia enables operational strategies for temporal flexibility and short-term demand response. A novel conic formulation incorporating this variable captures the dynamic behavior of smart control strategies in large-scale energy system optimization models, enabling up to a 22% reduction in heating-related electricity costs. The approach promotes substantial untapped flexibility potential through coordinated heat pump control or aggregator-based strategies.

A new framework for explainable artificial intelligence (XAI) has been developed to address the lack of transparency in AI/ML models used in network operations, using a large language model to generate human-understandable natural language explanations that can be translated into actionable insights, achieving 97.5% correctness and outperforming a state-of-the-art baseline.