In this paper, I discuss the connection between the structural properties of complex networks and consensus performance. Several experiments are performed to demonstrate how the structural properties and consensus performance change when the size of the network is changing. Three different network models, the square lattice network, the small-world network and the scale-free network, are examined by the experiments. A linear consensus model is adopted in this paper. Simulation results are provided to show theoretical prediction. It turns out that the square lattice network has the highest diameter and average path length and the lowest algebraic connectivity. The small-world network and scale free network have the ability to stabilize their structural properties when the size of the network becomes larger. The consensus simulation shows that the information propagation in square lattice networks is substantially inefficient. The system with 49 vertices approaches the consensus state exponentially with a convergence slope -0.1883. The convergence slope becomes -0.0111 when the vertex number of the network grows to 1000. On the other hand it is significantly efficient in small-world networks and scale-free networks. For both types of networks with 1000 vertices, the convergence slope stays around -0.6036 which indicates a fast decay rate. One extended experiment shows that adding edges to an existing network can significantly alter its structural characters and consensus convergence pattern
