DEVELOP OF SCHEDULING MODELS AND ALGORITHMS MTSSDRC UNRELATED PARALLEL MACHINE CONSIDERING TARDINESS AND WORKLOAD BALANCE

• Main Author: Amanda Larasati, Karina
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/84099
• Institution: Institut Teknologi Bandung

The proposed study pertains to scheduling of unrelated parallel machines with multi- task simultaneous supervision dual resource constrained (MTSSDRC), which considers the minimization of tardiness and workload balance. Tardiness is defined as the amount of completion time that exceeds the due date. Workload balance is calculated using the workload smoothness index (WSI). This research was developed because the scheduling of unrelated parallel machines better represents the real industrial conditions in the real world. The formulation of the MTSSDRC unrelated parallel machine scheduling problem for minimizing tardiness and WSI is a mixed-integer quadratically constrained programming (MIQCP) model. Finding solutions for the MIQCP model to simultaneously minimize tardiness and WSI is limited by the capabilities of available solvers. Therefore, an analytical solution to the MIQCP model is sought step by step. The problem-solving sequence begins with mixed-integer linear programming (MILP), which is used to solve the first objective function, tardiness, with additional solutions generated from the second objective function, WSI. Next, a mixed-integer quadratic problem (MIQP) is used to solve the second objective function, workload smoothness index (WSI), with the first objective function as a constraint based on the MILP solution. Finally, MIQCP is used to solve the first objective function, tardiness, adjusted by imposing a limit on the WSI value. This model yields a low WSI value within the specified WSI limits while adjusting the total tardiness. This research develops a metaheuristic NSGA-II algorithm to address the shortcomings of solvers in finding solutions for simultaneously minimizing tardiness and WSI. The NSGA-II algorithm is also capable of overcoming the limitations of solvers in problems involving many machines and operators. The development of Decoding Schemes 1, 2, and 3 is used to compare the experimental results produced by the different characteristics of these decoding schemes. These schemes represent the outcomes of both tardiness and WSI objectives, which will later be recommended to scheduling policy makers in the industry. Ratio y is used to group cases based on the comparison between operators and machines. This is done to observe whether there are differences in the solutions produced when these cases are grouped. The experimental results show that, in large cases, the two decoding schemes recommended for the industry with a ratio y = 0.5 are Scheme D2 and D3. Then, for a ratio y > 0.5, it is recommended that Scheme D1 and D2 be used in the industry as they can produce the smallest tardiness.