Types of Industrial Alarms (And Why Standing Alarms Are the Hardest to Track)
- Bruce Nicolson
- Jul 29
- 5 min read
By Bruce Nicolson, Principal Control Systems Engineer, Intelligent Plant. Bruce has 40 years of experience in the energy sector working on control systems and associated software.
Within a control room, an operator can expect to handle various types of alarms. Some flash up and clear within seconds, others remain active for hours or even days without requiring any action, and some occur so frequently that operators learn to just silence them without thinking. Understanding the different types of alarms, and why each one behaves the way it does, is the first step towards getting an alarm system properly under control.
Standing/stale, fleeting and chattering alarms
EEMUA 191, the industry guideline most commonly referenced for alarm system design and performance in the UK and Europe, provides a useful starting vocabulary.
The first thing to note is the definition of an alarm:
"An alarm is an audible and/or visible means of indicating to the operator an equipment malfunction, process deviation, or abnormal condition requiring a timely response."
Standing and Stale Alarms
An alarm is described as standing, or sometimes raised, for as long as the underlying condition persists. The two terms are used interchangeably. A stale alarm is simply a standing alarm that has remained active for an extended period, commonly defined as 24 hours. These typically occur on decommissioned plant, or when one half of a redundant pair is offline, which is often the case for pumps and compressors. Neither of these scenarios require a timely response from the operator, therefore fail the criteria of being an alarm.
Fleeting Alarms
Fleeting alarms are those that annunciate and then clear almost immediately, and do not return for a significant period of time, typically 10 minutes. As the alarm clears quickly by itself, it again fails the criteria of requiring a timely response.
Chattering Alarms
A chattering alarm as one that raises and clears at least three times a minute. This is one of the most common forms of nuisance alarm, and it's more significant than people often assume. It's not unusual for these repeating alarms to account for around half of all alarm annunciations at a typical plant.
Why fleeting and chattering alarms are such a problem
Fleeting and chattering alarms tend to be generated in one of a few ways: noise on a process variable when it's sitting close to an alarm setpoint, genuine high-frequency fluctuations in the process itself, or the repeated on/off action of a control loop.
Whatever the cause, the effect on operators is the same, and it's a nuisance for two reasons. Firstly, the operator has to silence or accept the same alarm every single time it occurs, which is tedious at best and distracting at worst. Secondly, in some alarm systems, these repeating alarms fill up the active alarm list with the same entries over and over, making it difficult to locate anything else that might be happening at the same time, including messages that actually matter. This becomes particularly dangerous during a plant upset, a key situation when an operator most needs a clear, uncluttered view of what's going on.
What good alarm performance looks like
EEMUA 191 provides a widely used benchmark for how often alarms should reasonably occur, based on the average rate an operator can be expected to handle without becoming overwhelmed.
Alarm Rate | Number of alarms per hour | Number of alarms per 10 minutes | Assessment |
>1 alarm /minute | >60 | >10 | Very likely unacceptable |
1 alarm/ 2 minutes | 30 | 5 | Likely over-demanding |
1 alarm/ 5 minutes | 12 | 2 | Manageable |
<1 alarm/ 10 minutes | <6 | <1 | Very likely acceptable |
The three KPIs everyone tracks, and the one that gets missed
Alongside these rate benchmarks, there are three core KPIs typically used to judge alarm system performance: the number of alarms per ten minutes, the peak number of alarms, and the number of standing alarms.
The first two of these are fairly straightforward to produce. They're essentially arithmetic performed on data that most historians and alarm systems already log in a structured way: alarms raised within a given window, and the busiest window across a period. Any reasonably capable system can generate these numbers directly.
The number of standing alarms is a different kind of problem entirely. It isn't a matter of counting discrete events, it's a matter of identifying a state that persists over time. To do that reliably, a system needs to work out, for every single alarm, the gap between when it was raised and when it actually cleared, and then filter for the ones where that gap crosses whatever threshold counts as standing or stale. Most historians and alarm event logs simply aren't built with that kind of query in mind. They're optimised to record that an alarm raised, that it was acknowledged, and that it cleared, not to answer the question of how long it sat there in between.
The practical result is that standing alarms tend to get lost in the sea of other alarm and event messages. Unless someone is actively searching for them, and manually tallying up the results, they stay effectively invisible. That's a real problem, because standing alarms are one of the more revealing indicators of chronic issues on a plant, whether that's a piece of degraded equipment awaiting attention or an alarm setpoint that was never right for how the process actually behaves in the first place. It's also, not coincidentally, an area that comes up regularly in improvement notices, precisely because it's the KPI most likely to have quietly gone unmeasured.
Ironically, it is an easy check for an inspector to make while actually at site; the control system alarm summary will directly display the number of alarms currently standing within the summary. It is not often that this display is available remotely.
How Intelligent Plant helps close that gap
Intelligent Plant's Alarm Analysis actively tracks the current status of alarms by monitoring the Alarm and Return messages, maintaining an internal state record, and writing the "On" time back into its database.
This can then be queried, and a report compiled of the alarms that have been on for a continuous (NOT clock) 24 hours within the requested report period. They are listed in order of greatest time on. Furthermore, this is a record of ALL alarms that have been standing over the period, not just those ones that were active when the control system was checked.
The report also details how many times the alarm annunciated, and how many of these were more than 24 hours.
The "barcode" display also shows periods of standing alarms, and if they are related:
Closing thoughts
Most of the alarm types described here, chattering, fleeting, and rate-based flooding, are relatively easy to spot once you know what you're looking for. Standing alarms are the exception. They don't announce themselves the way a flood or a chattering alarm does; they simply persist, quietly, until someone goes looking. If you're not sure how many standing alarms are currently active on your own system, that's usually the first, and most revealing, question worth asking.
🔗Check out related blog posts:
Comments