Using eye-tracking to investigate strategy and performance of expert and novice control room operators

There is lacking research within the Petrochemical Industry that uses eye-tracking to explore the differences between the strategies of expert and novice control room operators as they monitor and address process parameters that could be used to improve novice training programs and interface design. Scan paths and three eye-tracking metrics (Fixation Frequency, Gaze Duration Mean, and Gaze Percentage) were used to investigate the differences in eye behavior of three expert control room operators and six novice students as they monitored and corrected a Crude Refinement simulation. A 2x2x2 mixed factor design was used to explore the effects that expertise (expert and novice), interface type (black and grey), and alarm activity (active and inactive) had on participant eye behavior specifically, fixation frequency, gaze duration mean, and gaze percentage for certain areas of interest. The display was separated into 6 different areas and each area resulted in distinct eye statistics. Scan paths were plotted surrounding a subtle setpoint change within the simulation and were qualitatively analyzed to reveal differences due to expertise, interface type and alarm activity. The MANOVA revealed no significant differences due to expertise, interface type, and alarm activity. The single ANOVAs revealed that participants had higher fixation frequencies on the Main display during monitoring periods than during active periods revealing that both expert and novice participants’ attention was more divided when there were failures and alarms present than when the process was running at normal conditions. Also, experts spent a larger percentage of time monitoring the critical crude temperature and flow controller than novices. Pearson’s correlation between dependent variables revealed a positive correlation between fixation frequency and gaze percentage that indicated that participants typically had many, quick fixations rather than few, long fixations. Scan path analysis revealed that active monitoring and interface background color influenced how quickly operators discovered the setpoint change on screen. Overall, eye-tracking successfully detected differences between participants and interface types that can benefit novice training and display design