A genetic algorithm approach for balancing two-sided assembly lines with setups


ASSEMBLY AUTOMATION, vol.39, no.5, pp.827-839, 2019 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 39 Issue: 5
  • Publication Date: 2019
  • Doi Number: 10.1108/aa-11-2018-0192
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Page Numbers: pp.827-839
  • Erciyes University Affiliated: Yes


Purpose This paper aims to discuss the sequence-dependent forward setup time (FST) and backward setup time (BST) consideration for the first time in two-sided assembly lines. Sequence-dependent FST and BST values must be considered to compute all of the operational times of each station. Thus, more realistic results can be obtained for real-life situations with this new two-sided assembly line balancing (ALB) problem with setups consideration. The goal is to obtain the most suitable solution with the least number of mated stations and total stations. Design/methodology/approach The complex structure it possesses has led to the use of certain assumptions in most of the studies in the ALB literature. In many of them, setup times have been neglected or considered superficially. In the real-life assembly process, potential setup configurations may exist between each successive task and between each successive cycle. When two tasks are in the same cycle, the setup time required (forward setup) may be different from the setup time required if the same two tasks are in consecutive cycles (backward setup). Findings Algorithm steps have been studied in detail on a sample solution. Using the proposed algorithm, the literature test problems are solved and the algorithm efficiency is revealed. The results of the experiments revealed that the proposed approach finds promising results. Originality/value The sequence-dependent FST and BST consideration is applied in a two-sided assembly line approach for the first time. A genetic algorithm (GA)-based algorithm with ten different heuristic rules was used in this proposed model.