Crowdfunding (CF), an alternative source of financing for startups and creative projects, has gained prominent attention from practitioners and researchers. However, we find few research papers investigating CF from the perspective of its dynamic processes. In this paper, we propose a dynamic system framework, which incorporates heterogeneous entrepreneurs and funders, and CF platforms. In particular, we formally formulate the interactions among various factors in the CF processes. Through simulations we show that first, CF platforms are more viable for projects with small-scale capital requirements than those with large-scale requirements. Second, due to information asymmetry, non-profit projects are likely to acquire more funding comparing with for-profit projects. Third, CF platforms can reach maximum performance (indicated by the success rate of projects) by applying certain control mechanisms, such as acceptance rate and diffusion density. Our results are consistent with and provide a new and comprehensive angle for understanding the empirical results from current literature. We also discuss important practical insights of our research in details.