Categorization errors for data entry in maintenancework-orders

Abstract

Manufacturing is seeing a significant push toward digitization of processes and decision making. This push is enabled by the increased availability of data. Yet the work of the maintenance team, one of the core subsystems in any production line, remains a largely human endeavor. It involves manual work on the equipment and data collection by maintainers, which itself involves free-text and pre-specified categories or controlled vocabulary, rather than collection via sensors. Often this data is un-useful, in that it does not support the digitization of work. This paper presents an approach using Human Reliability Analysis (HRA) to identify the human errors associated with entering un-useful data. A Cognitive Task Analysis (CTA) is created, based on deconstructing the individual actions and decisions performed by maintainers in this process, backed by cognitive models to ground the task analysis in theory. A breakdown of human error modes for each task is provided as a list of Unsafe Acts (UAs), along with key contextual and organizational considerations, given as performance-shaping factors (PSFs). To demonstrate usage of this CTA, initial instantiation of a common HRA framework is provided as a case study, both to estimate base human error probabilities (BHEPs) and to motivate a discussion around initial risk mitigation strategies.