Abstract: There are many interdependent risk factors in the R&D projects. How to evaluate accurately these risk factors is an import basis for performing project risk monitoring and control. Using the Design Structure Matrix (DSM) and the method of ranking nodes in complex networks, this paper presents an improved K-shell method to perform quantitative risk analysis which can reflect the influence of complex risk network. First, we use the "risk event-risk factor" Multi-Domain Matrix (MDM) to deduce the dependencies among the risk factors, and build a combined propagation probability model. Further, by integrating the node's out-degree and in-degree centrality, the risk value of risk factors and propagation probability, we build the K-shell model for ranking nodes in complex risk networks. Finally, an industrial example is provided to illustrate the proposed model.

Key words: R&D project management, risk propagation network, K-shell method, Design Structure Matrix (DSM), risk analysis