Objective Pipelines are essential for the long-distance natural gas transportation and play a crucial role in national energy security. The linepack, as a comprehensive operational parameter for natural gas pipeline networks, embodies both safety and economic attributes. However, the existing steady-state linepack calculation method is only applicable under stable working conditions and lacks sufficient accuracy during variable operations, failing to address the unsteady characteristics of pipeline networks. Additionally, the dynamic linepack calculation method relies on complex online simulation systems, resulting in low calculation efficiency and challenges in meeting practical application needs.

Methods A rapid calculation method for linepack of natural gas was proposed by coupling data and mechanism models. First, the significance of the linepack index for the safe and efficient operation of the natural gas pipeline network was analyzed. Subsequently, based on pipeline industry standards and numerical simulation principles, a detailed comparison was made between traditional steady-state and dynamic linepack calculation methods, focusing on numerical accuracy, calculation efficiency, and applicable working conditions. To address the shortcomings of traditional methods in balancing accuracy and efficiency, a mechanism model was embedded within a data model and combined with Long Short-Term Memory (LSTM). This approach enabled efficient calculation of sequential linepack data, overcoming the poor generalization of pure data models and the low calculation efficiency of pure mechanism models.

Results The rapid linepack calculation method utilizing coupled data and mechanism models was applied to an in-service natural gas pipeline in China. Three coupling modes for the LSTM and the numerical linepack calculation model, including correction, embedding, and integration, were compared. Results indicated that the maximum linepack calculation deviation for the embedding mode was 0.76×10⁴ m³, significantly lower than that of the correction and integration modes. The rapid calculation method based on the embedding mode demonstrated significant advantages over traditional linepack calculation methods in accuracy and speed. It eliminated the need for complex online simulation systems, facilitated modular and lightweight linepack deployment, and validated the effectiveness of the proposed approach.

Conclusion The rapid calculation method for dynamic linepack, with coupled data and mechanism models, enables quick and accurate monitoring of natural gas linepack changes. This provides essential data and technical support for the safe, stable operation of natural gas pipeline networks and refined linepack management, assisting pipeline network companies in detecting inlet-outlet imbalances and ensuring fair access to pipeline network infrastructure.