For a physical model of the approach navigation channel of Three Gorges Project(TGP), flow patterns around a non-submerged spur dike placed as a mountain in a long rectangular glass flume were experimentally investigated with and without “sucking-spouting” water respectively. Based on rigid lid assumption and Dynamic Smagorinsky Model, all these vortex flows around the spur dike were numerically simulated and analyzed, which probably affect the whole flow field and then probably lead to sedimentations for silt-laden two-phase flows. Meanwhile, silt-laden flows were also investigated experimentally. Both the secondary flow region and the silt sedimentations downstream of the spur dike decrease greatly with “sucking-spouting” water compared to those ones without “sucking-spouting” water. The Finite Volume Method (FVM) was used to discretize the governing equations together with a staggered grid system, where the second order difference is employed for the diffusion terms and the source terms while the upwind difference QUICK is used for the convection terms. The computational results are all in fairly good agreement with the experimental data. For a physical model of the approach navigation channel of Three Gorges Project (TGP), flow patterns around a non-submerged spur dike placed as a mountain in a long rectangular glass flume were experimentally investigated with and without “sucking-spouting” water respectively. Based on rigid lid assumption and Dynamic Smagorinsky Model, all of these vortex flows around the spur dike were numerically simulated and analyzed, which probably affects the whole flow field and then probably lead to sedimentations for silt-laden two-phase flows. Meanwhile, silt- Both the secondary flow region and the silt sedimentations downstream of the spur dike decrease greatly with “sucking-spouting” water compared to those ones without “sucking-spouting” water. The Finite Volume Method (FVM) was used to discretize the governing equations together with a staggered grid system, where the second order difference is employed for the diffusion terms and the source terms while the upwind difference QUICK is used for the convection terms. The computational results are all in fairly good agreement with the experimental data.