An adversarial approach for the mixed-model assembly line design with new product variants in production generations

Assembly lines typically operate for several decades. Process engineers reconfigure the lines several times during the product family's life cycle, whereas product families may change several times a year in response to sales and marketing demands. These reconfigurations are often expensive and inefficient if the line is not flexible enough. The current study explores the feasibility of creating a line that takes product evolution into account during the line's life cycle. We study a line where a worker/robot and equipment pieces required are located at each station. When a new product model replaces one of the current variants in the product family, the line reconfigures to produce different product models from the same family. Reconfiguration can re-assign some tasks and rearrange equipment and resource elements. We formulate a model that accounts for the uncertainty of the product family evolution and the market demand. We propose an adversarial approach for the robust optimization of the mixed-model assembly line design for the worst-case scenario from a scenario tree for the future product family requirements. We run computational experiments using benchmark data. The results demonstrate that the developed adversarial approach outperforms the classical methods from the literature in terms of CPU time and solution quality. Copyright (C) 2024 The Authors. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/)