Real-time coordination of hydrogen injection in gas networks: a novel predictive framework

Hydrogen is increasingly recognized as a cornerstone of the energy transition, offering a pathway to decarbonize hard-to-abate sectors and enhance flexibility in integrated energy systems. However, the absence of dedicated hydrogen transmission and distribution infrastructure poses a major barrier to its widespread adoption. In the short to medium term, blending hydrogen into existing natural gas networks represents a practical and cost-effective solution, enabling large-scale transport without the immediate need for new pipelines. This approach, however, introduces significant technical challenges: gas quality parameters such as Wobbe Index (WI), Specific Gravity (SG), and Higher Heating Value (HHV) must remain within strict regulatory limits to ensure compatibility with network components and end-user appliances. Furthermore, when multiple hydrogen injection points and variable renewable generation are involved, the system becomes highly dynamic, requiring advanced control strategies to guarantee safe operation and optimal performance. This work proposes an innovative Model Predictive Control (MPC) framework designed to optimize hydrogen in real time injection into natural gas networks. The architecture combines three sequential optimization steps to address the complexity of multi-energy systems: • Energy dispatch optimization using a Mixed-Integer Linear Programming (MILP) algorithm, which computes optimal energy flows and hydrogen generation over the prediction horizon to maximize avoided CO₂ emissions, assuming no constraints on injection. • Gas network optimization formulated as a Nonlinear Programming (NLP) problem, which determines the maximum feasible hydrogen injection while enforcing gas quality standards and tracking gas composition through a batch-based approach. • Final energy dispatch recalculation using a second MILP algorithm, which adjusts the system operation based on actual injection limits to ensure consistency and environmental benefits. The MPC integrates a batch tracking technique to monitor hydrogen and methane fractions throughout the network, preventing violations of SG, WI, and HHV limits even under changing natural gas composition and additional biomethane injections. This capability is essential for real-world applications, where gas composition can vary significantly due to upstream injections or blending strategies. The proposed framework was validated in a Model-in-the-Loop environment on a complex network featuring multiple hydrogen injection points, several end-users, and a biomethane injection node. The results demonstrate that the controller effectively maximizes hydrogen injection while maintaining compliance with all gas quality constraints. Hydrogen is injected primarily during periods of low electricity grid carbon intensity and high renewable availability, thereby minimizing overall CO₂ emissions. The controller dynamically adjusts injection rates and electrolyzer setpoints to account for variations in carrier gas composition caused by biomethane, ensuring that hydrogen concentration remains within allowable limits. This adaptive behavior highlights the robustness of the approach in managing sector coupling between electricity and gas systems. From a computational perspective, the optimization process achieves execution times suitable for real-time operation in medium-sized networks (typically under two minutes per control step), although occasional outliers in more complex scenarios indicate the need for further refinement. The architecture is flexible and can incorporate additional objectives, such as economic performance or multi-objective optimization, without altering its core structure. The proposed MPC framework represents a significant advancement in the operational management of hydrogen blending into natural gas networks. By combining predictive control with coordinated optimization algorithms, it enables safe, efficient, and environmentally beneficial integration of hydrogen into existing infrastructures. This approach supports the transition toward low-carbon energy systems by leveraging renewable resources and existing assets, paving the way for practical deployment in real-world applications. Future work will focus on improving computational efficiency, extending the methodology to larger and more complex networks, and integrating economic considerations alongside environmental objectives.