EEMUA 191 and ISA-18.2 put a manageable alarm load at about one alarm per 10 minutes for each operator.
Both were written for process control rooms, but the reasoning carries over to a cleanroom EMS or BMS. Well above that rate, operators acknowledge without reading, and the pressure excursion that needed someone at the door scrolls past with the rest.
The alarm log already holds the evidence, and few sites count it. A month of history is enough. Divide the alarms by operators and hours for each shift separately, since one operator on nights carries a load that the day shift shares.
Then sort the tags by count. A handful often carry most of the load, typically pressure alarms whose delay is shorter than a normal door opening or whose limit sits inside normal variation. Anything that has stayed active for more than a day goes on a separate list, because a standing alarm usually points to a limit, a sensor or a room condition nobody has fixed.
Rationalisation works through the busiest tags one at a time, and the test for each is whether the operator has a defined action. If there is none, the signal can still be logged and trended as an event, off the alarm screen.
On a new system, that test can go into the URS so the alarm list is cut down before anything is configured.
Topics




