With increasing popularity of social networking websites, more and more people use social networking as a replacement for in-person interaction. This thesis develops a simulation model, under real condition and situation, to make experiment on Facebook so as to simulate how friends in social network interact and investigate how the interactions affect the overall structure of social network. The experimental results show that (1) The distribution of the number of interactions between each pair of friends has power-law phenomenon. (2) When the frequency of wall post by users is lower, it will lead to more significant power-law phenomenon. (3) The overall structural properties of social network remain relatively stable, with little change in clustering and average shortest path length over time.