Assembly Line Balancing Using Digital Twin Simulation for Reducing Cycle-Time Imbalance in Human-Centric Manufacturing

Nensi Lapotulo (Politeknik Pariwisata Batam)
Irma Darmawati (Universitas Pendidikan Indonesia)
Sitti Rachmawati Yahya (Universitas Siber Asia)
Makmur Harun (Universitas Islam Mulia)

Journal of Industrial Engineering and Management
Vol 2, No 1 (2024)

Cycle-time variation can constrain the practical performance of human-centric assembly lines when individual workstations operate above or below the planning target. This study examines a digital-twin simulation workflow for assembly-line balancing using a time study of 24 task elements, pre-balance workstation records from eight stations, 20 algorithm scenarios, and 60 post-balance validation observations. The task elements total 769.3 s, and the pre-balance reference takt is 90.0 s. The mean observed station cycles range from 88.0 s at ST05 to 98.5 s at ST07, with ST03 also above the takt reference at 97.6 s. Across the simulation records, mean line efficiencies are 70.08% for manual balancing, 75.88% for ranked positional weight, 81.77% for COMSOAL, 86.33% for the genetic algorithm, and 92.25% for digital-twin simulation. The 60 validation observations have mean line efficiency of 94.06%, smoothness index of 4.42, daily output of 365.42 units, takt compliance of 97.57%, and work-in-process accumulation of 16.3 units. These findings are descriptive summaries of the supplied operating records, not causal estimates. The proposed workflow combines measured task data, scenario comparison, human review, and validation feedback to support transparent balancing decisions