Abstract: The parallel operation of multiple projects is the norm for more and more engineering construction enterprises. The combination management of multiple projects focuses on coordinating the interactive relationships between projects, which can avoid vicious competition for resources between projects, improve resource utilization efficiency, and generate collaborative effects for enterprises. However, the interaction between projects increases the difficulty with risk management, leading to risk diffusion issues. When risk diffusion occurs, the failure phenomenon can spread between projects through interactive relationships, leading to cascading project failures and nonlinear losses. To solve the risk diffusion problem in project portfolios, this study aims to construct a risk diffusion model that conforms to the characteristics of engineering project portfolios, as a research framework for risk diffusion problems.
We first consider the project interaction relationship and the logical relationship of risk factors. Based on the mutual influence and interaction effect between these two different risk diffusion relationships (project interaction relationship and risk factor logical relationship), we construct a two-layer dependent network model for engineering project portfolio risk diffusion, and analyze the complex heterogeneous relationship of engineering project portfolio risk diffusion. Secondly, the cascading failure theory is used to propose risk diffusion rules and describe the logical path of the risk diffusion process. Finally, through the numerical simulation of the risk diffusion process of the project portfolio, the characteristics of risk diffusion are analyzed to verify the effectiveness and applicability of the model.
The simulation results find that: In the engineering project portfolio, the bi-layer dependent network model can reflect the diffusion characteristics of risks in the dual-relationship carrier of the projects and the risk factors; compared with the single-layer network risk diffusion model, the model shows a certain difference in the risk diffusion effect, the risk diffusion shows a more severe cascading failure process in the engineering project portfolio, and the project portfolio has weaker risk diffusion resistance. In practice, enterprise managers can develop preventive measures based on the risk diffusion characteristics of the model, such as: (1)Protecting important nodes in advance. The important nodes of a network generally refer to nodes with high degree, and the failure of these nodes will quickly affect most nodes in the network, reducing network robustness. Therefore, these project or risk factor nodes need to be closely monitored and managed for resource tilt. (2)Optimizing the risk network structure. Within the network layer, the robustness of the project meta network is improved by establishing redundant interaction relationships, adjusting the interaction relationships of a single project, and avoiding one project completely relying on another project. Between network layers, it is necessary to focus on controlling risk element nodes and project element nodes with many dependent edges, and weaken the interdependence between network layers in advance to suppress risk diffusion across network layers. (3)Backing up interaction relationships. For project interactions, project managers can design backup plans in advance from structural or functional aspects to prevent negative impacts on their own projects after the failure of interacting projects. However, the backup strategy requires an increase in cost and time, and in practical applications, it is necessary to consider the balance between the expenditure and return of this strategy. This model has practical significance for designing a more robust project interaction network for enterprises and improving their ability to cope with risk diffusion.
The risk diffusion model serves as a prerequisite for the risk diffusion problem in engineering project portfolios, and in future work, based on this risk diffusion model, network robustness analysis for risk diffusion will be further improved and applied to study network system stability, security, and other issues. This can provide management suggestions for enterprises to prevent and respond to risk diffusion, and provide new ideas and inspiration for further research on the stability of risk diffusion networks.

Key words: engineering project portfolio, risk diffusion model, bi-layer dependent network, cascading failure

摘要： 为构建符合工程项目组合特点的风险扩散模型,解决风险扩散问题。本文考虑风险扩散的双关系载体—项目交互关系和风险因素逻辑关系,构建工程项目组合风险扩散的双层相依网络模型,运用级联失效理论提出风险扩散规则,通过数值仿真分析风险扩散特征。仿真结果发现:工程项目组合中,双层网络模型能够反映风险在项目和风险因素双关系载体中的扩散特征;与单层网络风险扩散模型相比,构建的模型展现出差异性的风险扩散效应,风险扩散在工程项目组合中表现出更剧烈的级联失效过程。风险扩散模型作为工程项目组合风险扩散问题的前置性研究,为进一步风险扩散网络稳定性的研究提供新思路和启发。

关键词: 工程项目组合, 风险扩散模型, 双层相依网络, 级联失效