Sequencing multi-mixed-model assembly lines: An approach via clustering search

Abstract

Since lean manufacturing concepts have been adopted, several studies dealing with the effective utilization of Mixed-Model Assembly Lines (MMAL) have focused on the sequencing of such lines. The MMAL must have flexibility to produce different models in given sequences and obtain benefits, such as constant consumption of parts or subassemblies, thus minimizing the scaling of Kanban, the intermediate stocks, and the workload level at each station to minimize line stoppages. In situations where it is possible to produce many different models, production based on market forecast becomes unviable, even with the use of computational resources, which makes the products' sequencing in the MMAL a differential. This paper deals with the MMAL in multiple lines in a lean manufacturing environment, where an operational structure of several domestic suppliers supports many MMAL simultaneously, so that all the assembly lines can receive parts or sub-assembly from all the suppliers. To optimize this system, the sequencing must seek to minimize the distance between the real consumption and the constant ideal consumption of parts or subassemblies, thereby reducing the scaling of Kanban and intermediate stocks. To solve the sequencing problems, the Clustering Search (CS) was applied. Instances from the literature and also generated instances were tested, thus allowing to comparing the method with other methods presented in the literature. Analysing the results obtained, it was observed that the CS was efficient, obtaining good solutions in less time.