Stop treating alarms like advice

Spent my night chasing a low-flow alarm on a cooling loop that ping-ponged every 30 seconds. Everyone wanted a dashboard, I wanted a plug and a wrench. We found a partially collapsed suction hose and a clogged Y-strainer. The HMI logged 400 events. The pump logged 2 trips. Neither protected the asset. Well, there's your problem.

If an alarm requires an SOP, a reminder, and a hero, it is not an alarm problem. It is a design problem. Interlock it. Put a real differential pressure switch across the strainer, add a start permissive on minimum NPSHa, and give the pump a clean shutdown on low suction for 10 seconds, not 1. Add deadband and latching where it matters. Then you stop training operators to ignore noise.

We burn hours triaging nuisance faults instead of fixing the two-mechanical-things-per-line that always fail: loose mounts and dirty strainers. Anyone actually using ISA 18.2 style rationalization in brownfield plants, or do you just gut the alarm list during outage and move on? What simple interlocks have saved you the most grief?