One- and two-sided assembly line balancing problems with real-world constraints

In this study, we consider balancing problems of one- and two-sided assembly lines with real-world constraints like task or machine incompatibilities. First, we study the one-sided assembly line balancing problem (ALBP) with a limited number of machine types per workstation. Using a genetic algorithm (GA), we find optimal results for real-world instances. A set of larger test cases is used to compare two well-established solution approaches, namely GA and tabu search (TS). Additionally, we apply a specific differential evolution algorithm (DE), which has recently been proposed for the considered ALBP. Our computational results show that DE is clearly dominated by GA. Furthermore, we show that GA outperforms TS in terms of computational time, if capacity constraints are tight. Given the algorithm's computational performance as well as the fact that it can easily be adapted to additional constraints, we then use it to solve two-sided ALBP. Three types of constraints and two different objectives are considered. We outperform all previously published methods in terms of solution quality and computational time. Finally, we are the first to provide feasible test instances as well as benchmark results for fully constrained two-sided ALB.