In this paper a new heuristic for solving the assembly line re-balancing problem is presented. The method is based on the integration of a multi-attribute decision-making procedure, named “Technique for Order Preference by Similarity to Ideal Solution” (TOPSIS), and the well-known Kottas and Lau heuristic approach. The proposed methodology does not focus on the balancing of a new line, rather it takes into account the more interesting current industrial aspect of re-balancing an existing line, when some changes in the input parameters (i.e. product characteristics and cycle time) occur. Hence, the algorithm deals with the assembly line balancing problem by considering the minimization of two performance criteria: (i) the unit labour and expected unit incompletion costs, and (ii) tasks re-assignment. Particularly, the latter objective addresses the problem of keeping a high degree of similarity between previous and new balancing, in order to avoid costs related to tasks movements: operators training, product quality assurance, equipment installation and moving. To assess the performance of the presented approach a comparison with the original Kottas and Lau methodology is carried out. The results demonstrate the capability of the proposed algorithm of dealing with the multi-objective nature of the re-balancing problem. Solutions with advantages both in workload re-assignment, implying beneficial effects on the costs factors affected by tasks movements, and in completion costs are obtained in almost half of all problems solved. In the other cases, trade-off balancings with low increases in completion costs are presented.
The Assembly Line Balancing Problem (ALBP) consists in the assignment of tasks to operators engaged on the line in such a way that the final item is produced with respect to a pre-specified production rate. In literature, a wide variety of algorithms proposing to solve ALBP are found, however almost all of them consider this problem from a static standpoint, that is, before the line deployment.
Nevertheless, continuous changes in market requirements, regarding product design, restyling and quantity needed, combined with high product customization and reduced time-to-market tendency, highlight the need for managing dynamic versions of ALBP solution procedures. This need further increases when the paradigm of agile manufacturing, aiming at developing the capability to promptly respond to variations often unpredictable, is adopted. Particularly, in actual applications, the necessity of a re-balancing oriented design is great due to the occurrence of a wide variety of modifications in the input parameters, such as:
•
changes in product features, including tasks adding and removing, and variations in technological precedence relationships;
•
increase and decrease in performance tasks time, as a consequence of adopting new equipment;
•
modifications to the cycle time, due to changes in market demand.
Those modifications imply variations in tasks assigned to each station. When we deal with manual assembly lines, operators can learn how to perform the new requested tasks, but the learning process needs time and in the first phases involves errors. As a consequence, quality assurance and operators training, along with equipment switching costs, arise. Moreover, such costs are directly affected by the amount of changes in tasks re-assignment. To this extent, agility means capability to promptly face the modifications in the assembly process by minimizing tasks re-assignment, in such a way as to reduce the impact of the change. Hence, in this paper, an innovative heuristic algorithm for solving the assembly line re-balancing problem is presented, with the aim of minimizing both assembly cost and tasks re-assignment.
This paper addresses the assembly line re-balancing, a problem that nowadays frequently occurs in companies operating in competitive environments characterized by frequent changes in products features and sales volume. Such mutable scenarios affect assembly line balancing involving stations workload re-definition.
A multiple-objective heuristic procedure for solving the single-model stochastic assembly line re-balancing problem is proposed. Such an algorithm integrates the technique for order preference by similarity to ideal solution with the well-known Kottas and Lau heuristic approach.
The objectives taken into account are the minimization of the unit total expected completion cost along with the minimization of the tasks re-assignment. An index, called similarity factor, is introduced to measure the similarity between the task assignment in the initial and in the new line.
The results of a large-scale experimentation are given. In more than half of the problems solved, the proposed algorithm is able to provide solutions with significant improvements in both costs reduction and similarity increasing when compared to Kottas and Lau's methodology. In the other cases, the algorithm provides solutions characterized by a decrease either in assembly costs or in tasks re-assignment.
Particularly, the capability to reduce the re-assignment of operations involves positive advantages in reducing important costs factors, such as quality assurance, equipment switching, operators training, etc. Since these costs are difficult to estimate and are directly affected by the amount of re-assignments, the suitability of the approach based on the similarity factor is highlighted.
This work represents a valuable starting point for further studies centred on solving those assembly line balancing problems in which mutable sales scenarios constitute an important factor
