How many alarms are too many alarms? Alarm standards provide some guidance, and this study put those standards to the test using experienced operators. We found that EEMUA is relatively conservative in its recommendations.
Dr. Craig Harvey extended the successful experiment on the effect of alarms rates on operator performance. Actual operators and pipeline controllers responded to alarms using a pipeline simulator for one-hour periods at rates of 20/25/30 alarms per 10 minutes. The same experiment performed was also done with only alarm rates of 10 and 20 alarms per 10 minutes. Professional operator/controller performance did not degrade until 30 alarms per 10 minutes. Performance was the same as LSU students at 10 alarms per 10 minutes, but significantly faster (twice as fast) at the 20 alarm per 10 minute level.
Determine if variations in alarm rates and alarm presentation have the same effect on performance with operators/controllers as they did with college students.
- If and when does performance breakdown with increasing alarm rates for refinery operators and pipeline controllers?
- Does the breakdown differ for alarms presented by time of actuation (chronologically) or grouped by priority.
- How do results from student experiments compare to actual controllers/operators (e.g., do real operators/controllers perform the same as students)? The answer to this question will determine how future experimentation is accomplished (laboratory vs. field).
This project is an extension of the initial study evaluating alarm rates and interface style on operator performance. The initial study evaluated user performance two alarm display formats (chronological, priority) over 10-minute pipeline simulations with different alarm rates (1 per 10 min, 2 per 10 min, 5 per 10 min, 10 per 10 min, 20 per 10 min). The results indicated that a categorical display resulted in better performance and an alarm rate of 20 per 10 minutes resulted in significantly worse performance. However, the subjects were students and they only had to deal with the alarm rates in 10-minute increments of time. Therefore, this follow-on project was created in which actual operators/controllers would experience the alarm rates for long periods (60-minute increments).
- Comparison of the Experiment 1 results to the earlier student participant results.
- Operator performance (acknowledgment time, response time, accuracy of response, and successful completion, etc.) as a function of time and alarm rate.
- Buddaraju, D., Harvey, C. M., and Knapp, G. (2011). Performance of control room operators in alarm management, 2011 Proceedings IIE Research Conference.
- Thesis of Dileep Buddaraju
- Journal article on Glen Uhack’s work
- Journal article on Dileep’s Buddaraju’s