Learning and forgetting interactions within a collaborative human-centric manufacturing network

Abstract

Learning and forgetting (LaF) phenomena are characteristic of labor-intensive production and service industries. To mitigate the effects of LaF in a human-centric manufacturing system integrated with outsourcing, managers need to coordinate their decisions with partners for assigning operations and scheduling processes following a hierarchy. A model that addresses this should consider the expected latency of various tasks across assignments and production sequences and similarities among jobs as that affects learning. This paper develops a novel bi-level LaF model to help determine the leader-follower decisions in a decentralized network. It models the learning concept as a factor of task execution order and task variety. The mixed-integer non-linear optimization model determines the best order coordination and scheduling scheme by minimizing the processing, operating, and holding costs and penalties for missing deadlines. This study also develops an efficient column-and-constraint generation algorithm based on the duplication method, which enables solving bi-level models in which the lower-level model includes integer variables. This study also provides an illustrative real-sized example to validate the model and prove the efficiency of our resolution method. The results indicate that adopting compromise solutions enables preoccupied workers to be released earlier than expected, reducing the costs associated with learning and forgetting (due to latency). Despite the effects of LaF and the decentralized structure of the supply chain, which includes rising network costs, the schedules become more precise, and the cost balance among actors effectively increases.