Modelling Failure Propagation in Material-Driven Collaborative Manufacturing Networks Using Cellular Automata

Collaborative manufacturing networks enhance production efficiency but are increasingly vulnerable to cascading failures due to their complex interdependencies, particularly in critical processes like gear manufacturing. This study addresses this challenge by proposing a dynamic modelling framework based on Cellular Automata. Utilizing manufacturing resource and task scheduling data, a material flow-driven Directed Acyclic Graph (DAG) is constructed to capture the network’s hierarchical topology. Key innovations include state transition rules with memory effects, where dynamic failure probability integrates neighbouring node states and historical failure records, governing normal node failure, recovery, and re-failure (with an attenuation factor reflecting enhanced resilience). The case study focusing on the gear manufacturing industry, through simulations on a 100-node gear production network, reveals spatiotemporal failure propagation patterns. By implementing resource redundancy configuration and material flow optimization, iterations generally converge around 35 steps, demonstrating significant self-recovery potential and strong network robustness in collaborative manufacturing networks. This approach provides a scientifically grounded tool for identifying cascading risks in collaborative manufacturing networks.