Predicting the production quality risk of individuals in manufacturing: from empirical experiments to a predictive model

Abstract

Purpose – In this study we conduct experiments to quantitatively compare the relationship between human fatigue and quality in manufacturing assembly tasks. Our ultimate goal is to construct a model to predict the production quality risk of a human individual. Design/methodology/approach – We developed a VR simulation in which subjects perform common assembly tasks. We measure the tasks dimensional error as the quality indicator and time-of-day, task complexity, task duration, previous workload, etc. as fatigue factors. Findings – We observe known phenomena such as (1) learning reduces the task time for untrained subjects, and (2) poor ergonomic placement increased risk to quality. Moreover we observed a strong correlation between quality risk and time-of-day. The effects of task complexity are not so obvious and we are yet to decipher them. Research limitations/implications – The model developed can be used to optimize job schedules to maximize quality. Particularly in the case of jobs with varying duration, complexity, and quality requirements. This model is limited to individual human operators, and should be integrated in a simulation predict system wide behavior. Due to the nature of the virtual reality we cannot investigate the effects of physical fatigue effectively.