As an effective way to facilitate the increasing demand for reliable infrastructure, energy supply and sustainable urban development, underground utility tunnels have been developed rapidly in recent years. Due to the widespread distribution of utility tunnels, the safe operation of natural gas pipelines accommodated in utility tunnels has caused great concern considering fire, explosion, and other coupling consequences induced by the gas pipeline leakage. However, the limited information on leakage source terms in accidental leakage scenarios could preclude timely consequence assessment and effective emergency response. In this study, a BI-IEnKF coupling source term estimation (STE) model is developed, with the combination of gas dispersion model, Bayesian inference (BI) and iterative ensemble Kalman filter (IEnKF) method, to achieve the effective source term estimation (including leakage location and leakage rate) and gas concentration distribution prediction. The newly developed model is first evaluated by the twin experiment with good reliability and accuracy. Furthermore, three contributing factors affecting the performance of the developed BI-IEnKF coupling STE model were investigated to assist parameter selection for practical use. Additionally, the novel application of mobile sensors serving as an alternative for fixed sensors is explored, and an application framework is sequentially given to guide the deployment of the developed coupling model in utility tunnels. The results show that the developed model has great performance in accuracy, efficiency and robustness, as well as the potential to be applied in actual utility tunnel scenarios. This study can provide technical supports for safety control and emergency response in the case of natural gas pipeline leakage accidents in utility tunnels. Also, it could be helpful to reasonable references for gas lekage monitoring system design.

作为促进对可靠基础设施、能源供应和可持续城市发展日益增长的需求的有效途径，地下综合管廊近年来发展迅速。由于综合管廊分布广泛，考虑到天然气管道泄漏引起的火灾、爆炸等耦合后果，综合管廊内的天然气管道安全运行备受关注。然而，事故泄漏场景中有关泄漏源术语的信息有限，可能会妨碍及时的后果评估和有效的应急响应。在这项研究中，开发了一个 BI-IEnKF 耦合源项估计 (STE) 模型，结合气体扩散模型、贝叶斯推理 (BI) 和迭代系综卡尔曼滤波器 (IEnKF) 方法，实现有效源项估计（包括泄漏位置和泄漏率）和气体浓度分布预测。新开发的模型首先通过孪生实验进行了评估，具有良好的可靠性和准确性。此外，研究了影响所开发的 BI-IEnKF 耦合 STE 模型性能的三个影响因素，以协助实际使用的参数选择。此外，探索了移动传感器作为固定传感器替代品的新应用，并依次给出了应用框架，以指导开发的耦合模型在公用事业隧道中的部署。结果表明，所建立的模型在准确性、效率和鲁棒性方面具有良好的表现，具有在实际综合管廊场景中应用的潜力。该研究可为综合管廊天然气管道泄漏事故的安全控制和应急处置提供技术支持。也可为气体泄漏监测系统设计提供合理参考。