Objective As a crucial component of the global energy supply system, natural gas has seen its application scale continuously expand amid the ongoing transition to clean energy. However, with the accelerated pace of pipeline installation and the increase in service life, safety concerns arising from pipeline leaks have become increasingly prominent. Accidents involving natural gas leaks can lead to severe fires and explosions, posing direct threats to the lives and property of nearby residents. Therefore, there is an urgent need for an effective method for scenario construction and concentration field prediction related to natural gas pipeline leaks to enhance emergency response capabilities for such leaks.

Methods Based on the pipeline fracture control test field, a real-scenario drill platform for high-pressure and large-diameter natural gas pipeline leakage was developed. This platform is capable of simulating complex leakage scenarios with a burial depth of no less than 1.5 meters, a pipe diameter of no less than 1,422 mm, and a pressure-bearing capacity of no less than 14 MPa. Using the real data obtained from this platform, a three-dimensional computational fluid dynamics (CFD) model for pipeline leaks was established in conjunction with the Brinkman equation, enabling high-precision simulation of the diffusion and evolution process of leaked gas from buried pipelines. On this basis, a 3D concentration field decoupling method based on the algebraic iterative reconstruction technique was proposed. By establishing the mapping relationship between two-dimensional monitoring data and three-dimensional spatial concentration distribution, the three-dimensional reconstruction of the field domain was realized, addressing the issue of insufficient generalization ability in traditional models.

Results Leveraging the data provided by the real-scenario drill platform, the typical leakage scenarios of buried natural gas pipelines were accurately reproduced. In various leakage scenarios, the gas diffusion range in vertically upward leaks was significantly larger than that in leaks in other directions. In the initial stage of leakage, the momentum of the high-speed jet dominated the diffusion process, resulting in the fastest diffusion rate in the horizontal direction. In the later stage of leakage, the intensified gravitational effect led to a more pronounced downward diffusion trend of the gas. When the gas diffusion reached a steady state, the jet direction was consistent with the buoyancy direction of natural gas, creating a synergistic acceleration effect. This caused the volume fraction of natural gas on the ground to be significantly higher than that in horizontal and downward leakage scenarios, verifying the dominant role of leakage direction in the diffusion range. To validate the effectiveness of the decoupling method, comparative results showed that: the relative error of the maximum diameter of the gas cloud obtained from simulation and decoupling was 13.54％, and the relative error of the height was 11.83％. Meanwhile, the relative error of the natural gas volume fraction at each monitoring point ranged from 6.49％ to 14.92％. All the above errors meet the error requirement of no more than 20.00％ in the field of emergency response, effectively verifying the accuracy of the decoupling method in 3D concentration field reconstruction.

Conclusion The combination of the real-scenario platform and the decoupling algorithm addresses the accuracy challenges in scenario construction and 3D concentration field reconstruction for high-pressure and large-diameter pipeline leaks. It can provide key data support for planning the entry routes of emergency response equipment and selecting excavation methods in leak emergency disposal. This significantly improves the reliability of leakage simulations and the scientificity of emergency decision-making, and offers crucial technical support for the safe operation, risk prevention and control, as well as emergency disposal of long-distance natural gas pipelines in China.