AI‑Supported Multiphysics System‑Level Energy System Modeling and Optimization

#artificial-intelligence #power #systems #modeling #optimization #greentransition

Introduction: Energy systems modeling and optimization is essential for policy and grid operations, yet traditional approaches often sacrifice physical accuracy for computational speed. By focusing primarily on electrical variables (power, voltage, energy), current common standard models neglect the vital thermal and electro-chemical dynamics of modern, flexible assets, leading to grid bottlenecks and economic inefficiencies.

To bridge the gap between physical reality and system-level planning, a rigorous bottom-up approach is proposed:

• Multi-Physics Integration: Incorporating complex thermal and chemical behaviors into asset representation to prevent "missed connections" and grid delays.
• Hybrid Methodology: Combining Artificial Intelligence (AI) with physics-based optimization to ensure reliability without the typical computational burden.

Content:

1. Multi-Physics Integration I: Thermal incorporation and control for alkaline electrolyzers for system-level models
2. Multi-Physics Integration II: Aging-aware AC Optimal Power Flow of grid assets management (cables and power transformers) to unlock grid capacity
3. Hybrid Methodology I: Learning Active Constraints for Bilevel Optimization: Evaluating Classification Metrics on Real-World Distribution System Data
4. Hybrid Methodology II: Surrogate Models in Power System Scheduling Problems: Learning Active Constraints via Graph Neural Networks

• Welcome by Assistant Professor Haris Ziras
• Lingkang Jin: AI‑Supported Multiphysics System‑Level Energy System Modeling and Optimization
• Closing of event