Get your priorities (distribution) straight
A very common question is posed during alarm management training. Does the recommended alarm priority distribution of ~5% / ~15% / ~80% for high / medium / low priority alarms apply to the rationalized alarm priority distribution (as configured in the control system) or to the annunciatedalarm priority distribution (those alarms actually presented to the operator)? Is one priority assessment more important than another?
Alarm priority, which is the relative importance assigned to an alarm to indicate the urgency of response, is determined during the rationalization process. The effectiveness of alarm priority depends on the distribution of alarm priorities; higher priorities should be used less frequently (Ref ISA-18.2). If too many alarms are high priority, then effectively none of the alarms are high priority.
The ISA-18.2 / IEC 62682 standards provide recommended metrics for the annunciated alarm priority distribution as shown below:
The guidance in ISA-18.2 / IEC 62682 on rationalized priority distribution is much less precise.
“An effective alarm rationalization effort will produce an annunciated alarm priority distribution similar to the target distribution in the alarm philosophy. The annunciated alarm priority distribution will not match the rationalized alarm priority distribution since all alarms are not equally likely to occur.” (Ref 16.4.9, ISA-18.2)
This essentially says that a good rationalized priority distribution is one that leads to meeting the target annunciated priority distribution. There is no direct correlation between the rationalized and annunciated priority distribution, so defining rigid requirements for rationalized priority distribution is not useful. Instead good guidance for the rationalized priority distribution might be that there should be fewer high priority alarms than medium priority alarms, and fewer medium priority alarms than low priority alarms. Obviously, you are not likely to hit the target annunciated distribution for high priority (~5%) if > 50% of your rationalized alarms are high priority. Therefore, it is still important to review the rationalized priority distribution to assess whether rationalization was performed correctly.
Another interpretation of what is written in the standards is that the annunciated priority distribution is more important as a KPI than the rationalized priority distribution. Not all alarm management practitioners agree with this interpretation. Some believe that rationalized priority distribution is more important. To comply with the 5 / 15 / 80 % distribution targets, these practitioners assign a numerical score to each alarm during prioritization, where the score is calculated based on summation of different weighting factors (e.g. severity of consequences in environmental, in personnel safety…). The alarms are listed in order of score, and the 5% with the highest score are assigned to the high priority. The annunciated priority distribution is then accepted however it turns out.