Article Preview
TopIntroduction
The assembly line balancing (ALB) problem determines the workstations, which tasks are assigned to, by considering precedence constraints and cycle time so as to optimize predetermined performance measures (Ghosh & Gagnon, 1989). Many researchers have studied ALB problem since the first time it was introduced in the literature. Baybars (1986) classified ALB problems into two groups: simple assembly line balancing problem (SALBP) and generalized assembly line balancing problem (GALBP). In the SALBP, task times are considered as deterministic and a task can only be assigned to one station. While SALBP-1 aims to minimize the number of workstations for predetermined cycle time, SALBP-2 aims to minimize the cycle time for a given station number. GALBP signifies the any other ALB problem that generalizes SALBP by eliminating some assumptions. There are also many other classifications of ALB problems. The shape of the assembly line is one of the common classifying criteria for ALB problems. Miltenburg & Winjngaard (1994) and Erel, Sabuncuoğlu & Aksu (2001) studied U-line ALB problems. In addition to the single model assembly lines, mixed model (Yano & Bolat, 1989; Bard, Dar-El, & Shtub, 1992) and multi model (Burns & Daganzo, 1987; Dobson & Yano, 1994) assembly lines allow to produce similar or different products on the same line (Bukchin, Dar-El, & Rubinovitz, 2002). A detailed literature review can be found in Becker & School (2006) and School & Becker (2006). Additionally, some other specialized features such as parallel workstations (McMullen & Frazier, 1998), two sided lines (Bartholdi, 1993), parallel assembly lines (Gökçen, Ağpak, & Benzer, 2006), incompatibilities and bounded loads (Pastor & Corominas, 1999), designing assembly lines with equipment selection (Buckhin & Rubinovitz, 2003) and equipment costs (Buckhin & Tzur, 2000) are presented in the assembly line literature.
All these problem types arise from the goal of meeting the real-world problem requirements. Although the ALB literature is very rich, there is still a gap between theoretical studies and real-world applications (Becker & School, 2006; Boysen, Fliedner, & Scholl, 2008). Position and/or accessibility constraints (Lapierre & Luiz, 2004; Essafi, Delorme, Dolgui, & Guschinskaya, 2010) are considered as task assignment restrictions in addition to cycle time and precedence constraints. A certain station – task assignment may occur when the position of the assembling product cannot be changed due to weight or size of the product (School, Fliedner, & Boysen, 2010). Furthermore, zoning constraints can be applied when some group of tasks should be assigned to the same station or incompatible tasks should be assigned to different stations. All these constraints appear in real world applications, so these constraints should be taken into consideration while balancing the assembly lines. In this article, a real-world assembly line balancing problem is handled. An oven producer decided to redesign their assembly lines. They need to learn that whether it is possible to balance the assembly line in higher efficiency level while satisfying the production restrictions learned from previous assembly experiences.
ALB is an NP-hard problem (Karp, 1972) due to its computational complexity. Although technological developments allow us to handle different sized problems, it is still far away to deal with real world large sized problems. Gökçen et al. (2006) stated that the optimal solution of deterministic parallel ALBP was found for only small sized problems. Since the number of operations in real world ALB problems is very high, heuristic solution algorithms have been used to produce solutions within an acceptable time. The meta-heuristics have been widely used in many areas such as vehicle routing problem (Layeb, 2015), supply chain optimization (Gupta, Kundu, & Gupta, 2017) and other subjects are reviewed in Sensuse & Cahyaningsih (2018).