A pilot experiment for Science-based Human Reliability Analysis validation

Science-based Human Reliability Analysis (HRA) seeks to validate HRA methods through controlled experiments in nuclear power plant (NPP) simulators. This paper proposes a pilot experiment at The Ohio State University and introduces the necessary data collection tools. The experiment is built around the Information-Decision-Action Crew (IDAC) HRA method, a cognitive HRA method developed to model operator actions in accident scenarios. We focus on relationships between a select subset of Performance Shaping Factors (PSFs) that are observable in the NPP accident reports in the HERA database. The relationships between these PSFs are modeled in ID AC, and data collected will be used to validate this portion of the ID AC model. All independent PSFs are held constant except procedure quality, which is controlled at two levels; the dependent variables are the response PSFs: stress, global condition assessment, passive information load, time constraint load, task-related load, and attention. The proposed experiment compares operators responding to a Steam Generator Tube Rupture (SGTR) using two different sets of procedures. Data collection focuses on the response PSFs mentioned above. The data collection system includes operator self-assessments, expert observations, and physiological measures. Data collection instruments include questionnaires and scenario logs. Operators will be video-taped in the simulator as they respond to the SGTR, then they will watch a video of their own performance, pausing every minute to evaluate a variety of factors. The expert observer will watch the video and make similar assessments. Data collection instruments are introduced, and data analysis methods are proposed.