More Sensitive Alarms—Why Critical Exceptions Get Ignored

Why more sensitive alarms can make real exceptions easier to ignore

A device screen continuously displays temperature fluctuations, pressure deviations, communication reconnections, and maintenance alerts. The alarm sounds repeatedly, and the operator skillfully confirms, closes, and continues production. When a genuine alarm requiring shutdown inspection appears, it looks no different from the previous dozens of alerts.

Factory owners need to ask not about the total number of alarms, but which critical anomalies are being overlooked, who has the authority to stop production, and how quickly the shop floor can take the correct action.

An increase in alarm count is often interpreted as more comprehensive monitoring. However, human attention is limited, and if every alert competes for the same priority, the system trains operators to treat 'confirmation' as a screen-clearing action. The more sensitive the alarm, the lower the discernment may become.

The task of alarming is to promote action

An effective alarm should at least explain: what condition deviates, what impact it may have, what action the operator should take now, and how long before it escalates if unaddressed. Alerts lacking clear response guidance are closer to notifications and should not continuously occupy the highest level of attention.

A common practice on the shop floor is to set all risks as alarms, as designers fear missing any alerts. The result is the accumulation of low-value signals, forcing operators to develop their own filtering rules to maintain production.

This self-protective behavior is not unexpected. Confirming a false alarm incurs almost no cost, while actively stopping production may affect output and lead to accountability; when responsibility and authority are mismatched, continuing operations often becomes the safest choice.

If accepting orders proactively often leads to more errors, reducing workload can avoid accountability, making alarm responses difficult to handle truthfully. Response rules must clearly define authorization, risk boundaries, escalation, transfer, and closure responsibilities.

Priorities must align with different responses

Red should be reserved for statuses requiring immediate assessment or action, yellow for confirmation within a limited time, and ordinary alerts for tasks after the shift starts. Colors, sounds, and pop-ups should not only indicate severity but also link to the responsible person and time limits.

When the same root cause triggers ten linked alarms, the shop floor needs to see the initial signal and the impact scope, not ten repeated confirmations. After recovery, it is also necessary to determine which alarms disappear with the root cause and which still require individual handling.

Management should not only evaluate whether alarms are confirmed promptly. Quick confirmation might simply mean operators have learned to clear the screen faster. More valuable is the time from the first valid alarm to the correct action, as well as whether the same alarm repeatedly occurs.

First, remove alarms with no responsible party

Each week, select high-frequency alarms and ask four questions: who needs to see it, what action should be taken upon seeing it, is there enough time, and what is the closure standard. If answers are not available, downgrade to a record or add to the improvement list.

First, remove prompts with no action. Second, merge linked signals from the same root cause. Third, verify the responsible person and response time for each level.

Review workload again. The day shift has equipment, process, and quality support, while the night shift may only have operators and shift leaders. The same alarm strategy applied to both shifts results in different response capabilities.

Large process plants also involve strict safety regulations, making it impossible to arbitrarily delete or modify protective alarms. Discrete manufacturing sites cannot treat safety, equipment protection, and critical quality alarms as ordinary efficiency tools. Adjustments must retain risk assessment and approval boundaries.

Small, loosely managed factories should first distinguish between safety alarms and general alerts. Medium-sized growing factories should supplement with responsibilities and time limits. Large mature factories should focus on eliminating cascading alarms. Not all sites are suitable for using the same threshold settings.

Three signals your boss should see

First, which alarms occur most frequently but have never triggered action. Second, which critical alarms are always overwhelmed by other alerts. Third, which positions are required to respond but lack the authority to stop operations, isolate systems, or mobilize resources.

The sign of a truly mature alarm system is not how many changes can be captured on the screen, but whether the right people can take clear actions within a limited time when key changes occur. Fewer ineffective disputes will allow the shop floor to focus on the real red lights.

I am Huang Chaoxing, inventor of the TreeMES stack-light MES system. Follow me for continued analysis of shop-floor management and digital transformation trends in manufacturing.