MODELS FOR ASSEMBLY LINE BALANCING WITH HUMAN-ROBOT COLLABORATION AND ALTERNATIVE SUBGRAPHS

• Main Author: Cahyadi Nugraha, Raden
• Format: Dissertations
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/74653
• Institution: Institut Teknologi Bandung

The assembly line balancing problem (ALBP) plays an important role in producing an effective and efficient assembly production system. The main issue in ALBP is the assignment of a set of assembly tasks to a set of workstations in sequence without violating precedence constraints and other constraints to achieve a certain objective function. When the line cycle time as a representation of the production rate acts as the constraint, ALBP aims to minimize the number of work stations which means minimizing the cost. When the number of workstations is the constraint, ALBP aims to minimize the line cycle time, which means maximizing the production rate. Most assembly work is performed by human operators due to the complexity of the operations that require human’s dexterity and flexibility. However, humans also have limitations in terms of accuracy, consistency and strength. Human-robot collaboration (HRC) is an approach to overcome these limitations by utilizing the advantages of humans and the advantages of collaborative robots simultaneously. The use of HRC in assembly production systems promises high system performance in terms of productivity, quality, occupational health and safety, and flexibility. The use of HRC in assembly lines creates a new type of problem, namely assembly line balancing problem with human-robot collaboration (ALBP-HRC). Solutions to ALBP-HRC are urgently needed by several industries that have started using HRC in their assembly lines, such as the electronics industry and the automotive industry. ALBP-HRC is ALBP with the addition of decisions to determine whether a task is performed by humans only, by robots only, or by humans and robots together (HRC). The addition of this decision makes ALBP-HRC more complex. The use of both robots and HRCs can also provide alternative processes or routing to achieve a particular sub-assembly. In ALBP terminology, this collection of alternative processes is called an alternative subgraph. The existence of alternative subgraphs makes ALBP-HRC even more complex. This dissertation research aims to develop mathematical models for ALBP with HRC and alternative subgraphs (ALBP-HRC-AS). Thus, ALBP-HRC-AS is the problem of assigning a number of operations into a number of work stations, and at the same time determining whether the operation is done by humans only, by robots only, or by humans and robots simultaneously and involves the existence of alternative subgraphs on the precedence diagram. The novelty of this research can be viewed from two aspects, namely (1) the workstation takes into account the types of robotic tools (end effectors) to be used in each process that uses robots or HRC, and (2) the existence of alternative subgraphs. This research also considers two different conditions in the design of assembly trajectories, namely, first, conditions that allow the addition of resources, so the objective function is total cost minimization, and, second, conditions that do not allow the addition of resources, so the objective function is cycle time minimization. This dissertation research was conducted in four stages of model development. Stage A1 produced a basic model for ALBP-HRC that minimizes the total cost of humans, robots, and robotic tools. Stage A2 adapted the model generated in Stage A1 into a model that minimizes cycle time. Stage B1 further develops the model from Stage A1 to produce an ALBP-HRC-AS model that minimizes the total cost. Stage B2 adapts the model generated in Stage B1 into an ALBP-HRC-AS model that minimizes cycle time. The development of the mathematical model is done using a mixed-integer linear programming (MILP) approach. The optimal solution of the MILP model can be found by the exact method, but it requires a very long computational time, making it impractical to use for large problems. The developed mathematical model is able to generate optimal solutions to ALBP-HRC-AS, especially for small-sized problems. Meanwhile, for medium and large-sized problems, a metaheuristic algorithm based on ant colony optimization (ACO) was developed. The effectiveness of the developed ACO algorithm is shown by the ability to produce solutions with an average gap to the exact method solution of less than 10%. As an illustration of efficiency, the computational time required by the ACO algorithm, for large problems, is less than 10 minutes, compared to the exact method, which takes hours.