Interface management of complex capital projects from the perspective of interface clusters: Using complex networks and spectral clustering methods

In complex capital projects, the interfaces among stakeholders with diverse areas of expertise are often highly interdependent, resulting in substantial project risks. Previous studies, however, tended to analyze and manage these supposedly interdependent interfaces separately, thereby ignoring the potential propagation chain of interface risks generated by the dependency network. The failure to clearly identify and manage the chain may lead to exacerbated conflicts and tensions among stakeholders, ultimately resulting in project delays or failure. Therefore, the authors propose a method for interface clustering based on dependency strength between interfaces, aiming to maximize the interaction rate among project interfaces within clusters while minimizing it outside of clusters. The method effectively mitigates the propagation of interface risks and enhances communication and coordination efficiency between interface owners by consolidating highly interdependent interfaces within clusters. To achieve this goal, this paper first models the mutual dependence relationships and strengths of project interfaces using an interface dependency matrix and an interface dependency network. Subsequently, a spectral clustering algorithm is employed to divide interfaces into clusters with distinct local objectives based on their dependence relationships. Finally, by studying a rail transit project in China, the authors demonstrate the important role of the proposed method in reducing complexity in interface management, decreasing occurrence and spread of interface risks, and enhancing communication and collaboration among diverse stakeholders. The contribution of the method proposed here is to enrich existing interface management theories and offer project managers novel methodologies and theoretical support for formulating strategies to address interface issues.