We propose an evolutionary model for weighted network with tunable clustering coefficient according to characteristics of real network. The model gives power-law distributions of degree, weight, and strength, as confirmed in many real network. In particular, the weighted model has a nonlinear correlation between average clustering coefficient and degree, which is in good agreement with flat head real weighted technological network. Moreove, the effect of the weighted network structure on traffic delivery is studied. The packet traffic flow on the weighted scale-free network is investigated based on the local routing strategy using node strength, and the delivering ability of node is controlled by node strength. The simulations show that the traffic dynamics depends strongly on the controlled parameter.