Industrial assembly lines with multi-operated workstations: applications and methods

Abstract

Assembly lines are widely present in the automotive manufacturing industry. The procedure of building vehicles employs several workers or robots equipped with a diverse pool of tools. Facilities, wages, robots, and tools are quite costly, giving rise to the necessity of consciously designing an efficient line. It is crucial to meet product demand and reduce expenses at the same time. Due to large-size products found in automotive industries, multiple workers can be assigned to the same workstation in order to simultaneously perform different operations on the same product. In this PhD thesis, three proposed studies are presented and discussed: the Robotic Assembly Line Design (RALD) problem and two variants of the Multi-manned Assembly Line Balancing Problem (MALBP). Mixed-Integer Linear Programming (MILP) formulations are developed for all three problems. They either aim at minimising total costs at the desired production rate or the cycle time given limited resources. For the RALD problem, several practical characteristics are taken into consideration, computational tests are conducted, and a practical case study is solved with parameters collected from a real-world robotic welding assembly line, reaching optimality. Secondly, both variants of MALBP models incorporate strong symmetry break constraints (valid inequalities) and decompose the original problem into innovative Benders’ Decomposition Algorithms (BDA). These algorithms are able to optimally solve large-size instances and outperform previous methods in terms of solution quality. Finally, contributions of developed works are summarised and further research directions are suggested for all problems.