HMI alarms: classes, priorities and acknowledgement

Alarm, event, message

• Alarm — abnormal condition requiring a timely operator response.
• Event / message — information worth logging (equipment started, mode changed) but no action needed.

Mixing them is the main reason operators stop reading alarms.

Classes and priorities

HMI systems group alarms in classes (e.g. errors, warnings, system messages) which define acknowledgement model, colors and logging. Within process alarms, priority expresses urgency and consequence — typically 3–4 levels. As a rule of thumb the higher the priority, the fewer alarms should have it.

States and acknowledgement

Acknowledgement means "I have seen it". It does not reset the equipment: resetting a trip is a separate command, usually only possible when the cause is gone.

Presentation

• Alarm banner on every screen with the newest/highest-priority alarms.
• Alarm summary with filters by area and priority, time of activation and acknowledgement.
• Direct navigation from an alarm to its screen/faceplate.
• Alarm texts that say what, where and ideally what to do: "P-101 bearing temperature high (85 °C) — check cooling water".

Floods and shelving

During an upset hundreds of alarms can arrive in minutes (an alarm flood); operators then miss the important one. Remedies: rationalize alarms, state-based suppression in the PLC, first-out indication, and shelving (temporarily hiding a nuisance alarm with time limit and logging) — under change management, not ad hoc.