A computational fluid dynamics (CFD) model for natural gas pipeline leakage in an underground integrated utility tunnel was developed using double-precision ANSYS Fluent, incorporating the actual parameters of the Xinyong Road utility tunnel in Nanning. Leakage, intake, and exhaust conditions were configured, and the model's accuracy was validated by comparing it with experimental data. Quantitative analysis of gas diffusion behavior and concentration distribution was conducted, revealing that initially methane forms a high-concentration zone near the leakage point, with rapid upward diffusion in the tunnel within seconds. Over time, methane concentration becomes more uniform across the tunnel cross-section, with a noticeable dilution effect from 50 to 300 seconds, eventually achieving a steady distribution on a longer timescale. Monitoring data indicated that the highest concentration rise near the leak, stabilizing after 25 seconds, with intake points stabilizing around 1.5% and exhaust points below 0.1%. These findings provide quantitative insights into methane concentration distribution, diffusion rate, and spatial spread following leakage, highlighting the patterns of initial and long-term gas concentration change. This study supports the development of effective safety controls, optimized ventilation designs, and emergency response strategies.