Alarm Fatigue in Manufacturing: A Human Factors Perspective

When a critical alarm goes unacknowledged and an incident follows, the investigation often lands on the operator. They should have seen it. They should have responded faster. The language of alarm fatigue suggests the human failed.

Cognitive load research tells a different story. Human working memory can hold approximately four to seven items simultaneously (Cowan, 2010). When an operator is presented with alarm rates that exceed one every two minutes, a level EEMUA Publication 191 (4th Edition, 2024) classifies as “likely to be over-demanding,” the brain does not fail. It adapts. It begins filtering, prioritising by recency and familiarity rather than by configured priority. Critical alarms that look and sound like every other alarm get filtered out.

This is not negligence. It is a predictable cognitive response to an unsustainable information load. Your existing alarm rationalization programme addresses the alarm system side. This post addresses the human factors side: why operator cognition is the constraint, and what that means for system design.

What the Benchmarks Actually Represent

The EEMUA 191 benchmark of one alarm per operator per ten minutes during normal operations is not an arbitrary target. It represents the rate at which a single operator can reliably receive an alarm, assess the process context, determine the correct response, execute that response, and confirm the outcome before the next alarm arrives.

That ten-minute cycle assumes the operator is doing nothing else. In practice, operators are also monitoring process trends, communicating with field technicians, logging data, and managing routine tasks. The effective capacity for alarm response is lower than the benchmark suggests.

When alarm rates exceed the benchmark by a factor of five or ten, which is common in unrationalised plants, operators are not processing alarms individually. They are scanning, batch-acknowledging, and relying on pattern recognition. The alarms that break through are the ones that look different from the background, not necessarily the ones with the highest configured priority.

Three System-Level Interventions

Addressing alarm fatigue as an engineering problem rather than an operator discipline problem leads to three interventions.

First, rationalization reduces the total alarm load. ISA 18.2 and EEMUA 191 both prescribe a structured lifecycle for reviewing every configured alarm and removing or reconfiguring those that are not unique, timely, relevant, and actionable. Rationalization typically removes 30% to 60% of configured alarms (ISA, ANSI/ISA-18.2-2016).

Second, intelligent routing reduces the effective alarm load per operator. Instead of broadcasting all alarms to all operators, SeQent Alarm Management routes each alarm to the specific person responsible for that area and shift. An operator covering Zone A does not receive Zone B alarms. The cognitive load per operator drops without removing any alarms from the system.

Third, shelving provides a mechanism to temporarily suppress known alarms during maintenance, startup, or shutdown. This prevents expected alarms from contributing to operator workload during periods when they are anticipated and not actionable.

Measuring It Before You Fix It

You cannot reduce alarm fatigue without first quantifying it. The key metrics are: average alarm rate per operator per hour during normal operations; alarm flood frequency (periods exceeding ten alarms in ten minutes); standing alarm count (alarms active for more than 24 hours); and acknowledgement response time distribution.

SeQent Alarm Management provides the alarm performance data needed to calculate these metrics. The Alarm Health Scorecard at seqent.com/alarm-scorecard/ provides a quick baseline assessment.

Score your alarm system or contact the team to discuss alarm fatigue reduction in your plant.