Optimal trajectory-guided stochastic co-optimization for e-fuel system design and real-time operation

Researchers developed MasCOR, a machine-learning framework for optimizing e-fuel production systems that combines design and operational decisions under renewable energy uncertainty. The system demonstrates near-optimal performance with significantly lower computational costs than traditional mathematical programming approaches.

• →MasCOR uses machine learning to co-optimize e-fuel system design and real-time operations under renewable energy uncertainty.
• →The framework shows substantially lower computational costs compared to traditional mathematical programming while maintaining near-optimal performance.
• →Testing across four European sites revealed most locations benefit from reducing system load below 50 MW for carbon-neutral e-methanol production.
• →E-methanol production costs of 1.0-1.2 USD per kg were achieved across the tested European locations.
• →Dunkirk, France required different optimization strategies due to limited renewable availability and high grid prices.