Three-Dimensional Source Localization of Natural Gas Leakage Using an Experimentally-Calibrated Gaussian Plume Model and Enhanced Particle Swarm Optimization

Abstract To address the challenges associated with the three-dimensional source localization of surface natural gas micro-leakages, this study proposes a novel three-dimensional localization method that integrates an experimentally-calibrated Gaussian plume model with an enhanced particle swarm optimization algorithm (EGPM-EPSO). The proposed method was evaluated systematically through controlled test-rig experiments, followed by preliminary field demonstration. A well-conditioned experimental platform was first designed to quantitatively characterize the spatial distribution of the leaked gas concentrations using tunable diode laser absorption spectroscopy. A plume rise impact factor was introduced to describe the coupled effect of the leakage rate and wind speed, based on which an exponential saturation model for the plume rise height was developed. An experimentally-calibrated Gaussian plume model (EGPM) was established by further deriving the lateral and vertical diffusion parameters using experimental data. Compared to traditional models, the EGPM substantially improved the concentration prediction accuracy, reducing the average relative error to 1.65%. The EGPM was then integrated with an enhanced particle swarm optimization algorithm incorporating several strategic improvements, achieving a mean localization error of 0.0084 m on the experimental platform. Additionally, field testing for natural gas leakage was conducted via a vehicle-mounted mobile monitoring system. The results indicate that the proposed three-dimensional localization method can accurately identify the leakage source with a total localization error of 3.1 m. The measurement-driven inversion framework establishes a robust foundation for the safety monitoring and emergency response of gas pipeline leakages.