Robustness analysis of large scientific facilities development network with different cascading failure modes

Developing a large scientific facility involves numerous sub-products and research organizations, which pose systematic risks of potential global cascading failure. The development process can be abstracted as an interdependent network, the Large Scientific Facility Development (LSFD) network, consisting of sub-networks of product and organization. This paper offers a comprehensive risk cascading model by integrating the Load-capacity and Epidemic models. It explores the risk cascading mechanism and robustness of the LSFD network under diverse assumption conditions. The results show that the robustness is controlled by network structure, coupling patterns, attack strategy, node load & capacity, and probability of infected & removed. Enlarging the network size leads to a decrease in robustness, and increasing the average degree of the product network can improve the robustness. The assortative coupled networks exhibit the worst robustness when encountering an intention attack. Moreover, adjusting capacity parameters can significantly impact cascade failures, highlighting the inevitability of risk cascade even with sufficient resistance capacity. The removed probability λ determines whether the network faces nearly collapse: if λ ≥ 0.5, over 90 % of nodes will fail. These simulation results provide practical implications for managing systematic risks and enhancing the robustness of the LSFD network.