Assembly line balancing for personalized production

Assembly line balancing problems aim to an efficient and effective assignment of all the required tasks to workstations in a flow oriented production system. Nowadays, assembly lines have to face the manufacturing of extremely personalized products (e.g. cars) as requested by an increasingly higher portion of the market demand. Several literature contributions focus on different balancing problems affected by the wide variety of the final product, e.g. mixed and multi model assembly lines. However, no contribution seems to tackle the personalized production of goods. Such products require to assemble a certain number of tasks whatever the final product personalization is, and a variable number of optional of different type determined by the specifications of every single costumer. This paper faces the generalized assembly of personalized goods proposing an innovative two step methodology to optimize the workload balancing between the assembly line stations, considering traditional tasks and the optional required by the product personalization, which could occur with different frequencies and pairings. The first phase of the developed methodology executes a clustering of product options required by the costumers based on a similarity index. This phase leads to the definition of several sets of optional typically requested together by the customer and with similar mounting time. The methodology second phase leverages the defined clusters of optional. Indeed, optional of the same cluster shouldn't be assigned to the same workstation to reduce the overload or underload of the assembly operators. An integer programming model is proposed to assign both traditional tasks and optional to stations, to maximize the assembly line balancing considering the order frequency and assembly time of the clusterized optional. An industrial case study is adopted to test and validate the proposed two steps methodology. The obtained results highlight a consistent time balancing between assembly line workstations and a significant limitation of the operator overloads. Copyright (C) 2020 The Authors.