Disassembly line design with multi-manned workstations: a novel heuristic optimisation approach

Çevikcan E., Aslan D., YENİ F. B.

INTERNATIONAL JOURNAL OF PRODUCTION RESEARCH, vol.58, no.3, pp.649-670, 2020 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 58 Issue: 3
  • Publication Date: 2020
  • Doi Number: 10.1080/00207543.2019.1587190
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, ABI/INFORM, Aerospace Database, Business Source Elite, Business Source Premier, Communication Abstracts, Compendex, INSPEC, Metadex, zbMATH, Civil Engineering Abstracts
  • Page Numbers: pp.649-670
  • Keywords: disassembly, line balancing, multi-manned stations, remanufacturing, optimisation, constructive heuristic, BALANCING PROBLEM, COLONY ALGORITHM, SEARCH ALGORITHM, MODEL, FORMULATIONS, PRODUCT, WORKERS, AHP
  • Karadeniz Technical University Affiliated: Yes


As the first and the most time consuming step of product recovery, disassembly is described as the systematic separation of constituent parts from end-of-life products through a series of operations. In this context, designing and balancing disassembly lines are critical in terms of the efficiency of product recovery. Recent research on disassembly line balancing (DLB) has focused on classical stations where only one worker is allocated. However, such a line results in larger space requirement and longer disassembly lead time. In this paper, disassembly line balancing problem (DLBP) with multi-manned stations is introduced to the relevant literature as a solution to overcome these disadvantages. A mixed integer linear programming (MILP) model and two novel framework heuristic algorithms are developed to minimise the number of workers and workstations. MILP model has been applied to a dishwasher disassembly system. The application results indicate the superiority of establishing multi-manned stations over classical disassembly system design with single-worker stations with shorter disassembly lead time (80.9%) and line length (60.2%). Moreover, the proposed heuristics have been compared on newly generated test problems (instances) for DLBP. The results validate that the heuristics provide acceptable solutions in a reasonable amount of time even for large-sized problems.