A Bayesian network based framework for ripple effect analysis in cross-tier shipbuilding supply chains

Modern shipbuilding supply chains are increasingly vulnerable to cascading risks across multiple tiers. However, existing analytical models fail to adequately capture the nonlinear propagation of disruptions due to oversimplified dependency mappings and the neglect of latent risks. This limitation impairs managers’ ability to effectively manage ripple effects within multi-level production systems. This study proposes a Bayesian network (BN) with a leaky noisy-or to evaluate disruption propagation and resilience in both multi-level cross-tier and single-level linear shipbuilding supply chains. The framework models bidirectional dependencies and latent risks, quantifying how cross-tier structures leverage redundant pathways to mitigate downstream disruptions. Additionally, it identifies critical nodes and evaluates disruption scenarios through probabilistic analysis of interdependency. To validate the proposed framework, we conducted a 23-node shipbuilding supply network based on the operations data of J-shipyard. This work comparatively analyzes between cross-tier BN structures and conventional linear network representations. The cross-tier network demonstrated superior performance reducing downstream disruptions by 10.3%, thereby highlighting the resilience advantages of alternative pathways. Sensitivity analysis revealed that raw material suppliers are critical vulnerabilities, as indicated by a sensitivity index exceeding 0.25. This research advances the field of disruption analysis in complex production systems by enabling precise risk prioritization and optimizing pathway redundancy. Practitioners can utilize this framework to enhance the resilience of tiered supply chains, particularly in industries characterized by complex supply chains. The adaptability allows for its application across various multi-level organizations.