The dispersion of air pollution in complex situations such as the cases of the filled road is a significant problem for the public safety and living quality. Application of computational fluid dynamics (CFD) helps to build the model calculation in order to estimate the dispersion of air pollutants. In order to assess its accuracy, this study used the Realizable k-ε model, the RNG k-ε model, and the Shear-Stress Transport k-ω turbulence model in FLUENT CFD code. The results were compared with the wind tunnel experiments. The Realizable k-ε turbulence model provided the best prediction for the surface concentration and concentration profiles of selected downwind positions of the filled road. It was found that a noise barrier, which positioned on the filled road, increases the vertical air pollution impact distance larger 1.75~1.92 times and decrease the horizontal impact distance lower 0.46~0.54 times than those of no barrier case. It was also found that two or three noise barriers increase 1.63~1.79 times the vertical air pollution impact distance. It contributes the decrease of horizontal air pollution impact distance 0.49~0.63 times compare with no barrier case.