Assembly Line Balancing in an Automotive Cables Manufacturer Using a Genetic Algorithm Approach Hager Triki, Ahmed Mellouli, Wafik Hachicha, Faouzi Masmoudi Laboratory of Mechanic, Modeling and Production (L2MP) Engineering School of Sfax, Tunisia hager_triki@yahoo.fr Abstract—In general, assembly lines, allow low production costs, reduced cycle times and accurate quality levels. Assembly line balancing problem (ALBP) is about how to group the assembly activities, which have to be performed in an assembly task, into workstations, so that the total assembly time required at each workstation is approximately the same (cycle time). In this paper, an ALBP is studied in a real-world automotive cables manufacturer company. This company found it necessary to balance its line, since it needs to increase the production rate. In this line type, the number of stations is known and the objective is to minimize cycle time where both precedence and zoning constrains must be satisfied. For this purpose, a hybrid genetic algorithm (HGA) based on genetic algorithm scheme and local search procedures is implemented and fully characterized. The results of comparative studies with CPLEX software exact solutions show that the proposed HGA approach is very effective and competitive Keywords—Assembly line balancing problem; precedence and zoning constrains; cycle time; hybrid genetic algorithm approach; real-world case study I. INTRODUCTION An assembly line balancing problem (ALBP) consists of distributing the total workload for manufacturing products among the stations along the manufacturing line. Research has focused on developing effective and fast solution methods for solving the simple assembly line balancing problem (SALBP) [1] and their various extensions. Each extension is motivated by several real-life applications and the need for solving precise practical problems. In this paper, we study the process of assembly line balancing of motor cables in the multinational company LEONI (http://www.leoni.com ) to increase the production rate. The company is a global supplier of wires, optical fibbers, cables and cable systems as well as related development services for many applications in industries especially the automotive business. In this industry, we describe a balancing assembly line when zoning constrains exist among some pairs of tasks. For example, some raw materials (file, tube etc...) can be so similar to each other that quality consideration and the processing conditions force certain pairs of tasks to be assigned to different stations. The main characteristics of this ALB Problem (ALBP) are as follows: (a) the performance time of each operation is deterministic (b) the precedence relationship among assembly tasks is known and invariable (c) a simple product type is assembled on the line (d) no buffer is considered between the stations (e) The zoning constraints are included in tasks assignment to stations. In this situation, the number of stations is known and the objective is to minimize cycle time where both precedence and zoning constrains must be satisfied. All these constraints make this ALBP very hard to solve. In literature, there are various applications of metaheuristics to solve large scale of ALBPs. Among these metaheuristics, Genetic Algorithms (GAs), which is received an increasing attention from the researchers since it provides an alternative to traditional optimization techniques by using directed random searches to locate optimum solutions in complex landscapes [2]. GAs are a stochastic procedure which imitates the biological evolutionary process of genetic inheritance and the survival of the fittest. As a population-based approach, GA has two main advantages: • GA can find a population of solutions rather than a single solution in a single run. Consequently, the likelihood that the algorithm will be trapped in a local optimum decreases. • GA fitness function can take any form and several fitness functions can be used simultaneously [2] Therefore, we chose to apply the GA method because it is a powerful tool used to found good solutions for NP hard problems Although there are plenty of previous researches that use the algorithms, only the underlying and the related GA framework adopted in this paper is presented due to space limitation. Those readers who are further interested in GA may refer to [3] and [4]. The majority of these previous studies have validated their GA performances using simulate problem, but little via real case studies data. For example, [5] developed a GA to solve SMALB Type-1 problem in the clothing industry in order to maximize the line efficiency. Reference [6] reduced the cycle time by 28.5% of a real two sided car assembly line by applying a GA. However, these studies have not integrated an optimization tool to choose their AG parameters. Since the problem is NP-hard, a hybrid genetic algorithm (HGA) based on genetic algorithm scheme and local search procedures is implemented and fully characterized. The main steps of HGA are explained in the next section. International Conference on Advanced Logistics and Transport 978-1-4799-4839-0/14/$31.00 ©2014 IEEE 336