Stop Adding Alarms; Start Fixing Causes

Had a lovely 2 am chorus of LOW FLOW, LOW SUCTION, SEAL POT LOW until the pump finally ate itself. Root cause wasn’t exotic. Valve proving was disabled for flexibility, min-flow bypass cracked shut, and we let a chattery transmitter spam the HMI. Well, there’s your problem.

We ran a one-day cleanup: added time delays and hysteresis on noisy PVs, moved FYI items to advisories, latched true trips, and put starts behind hard permissives: suction > X, min-flow proved open, seal pot > Y. Also set maintenance overrides to expire. Result: alarm rate down 70%, zero nuisance trips this month, no seals sacrificed.

My rule of thumb: if no defined operator action in 60 seconds, it isn’t an alarm. If the right action is always the same, interlock it. Alarms protect people and equipment, not feelings.

How do you pick thresholds and delays for rotating equipment? ISA 18.2 by the book, or shop rules? Where do you draw the line between alarm and permissive?