Fixing Your Nuisance Alarms Just Got a LOT Easier

The most popular alarm management presentation we have ever done has been the one on fixing “Bad Actor” alarms. I like it because I get to refer to some of my favorite schlocky sci-fi movies – and it turns out that lots of engineers like those as well!
 
Nuisance alarm reduction is an important step. We recommend it at the beginning of an alarm improvement effort because it is easy, fast, cheap, and makes a BIG difference in alarm system performance. An early win is the best way to start!
 
In The Alarm Management Handbook, we devote an entire chapter to the methods of reducing several varieties of nuisance alarms. For chattering and fleeting alarms, the most powerful tool for reducing them is often the use of properly-selected alarm on-delay or off-delay settings. A brief review:
 
Fleeting alarm: a short-duration alarm that clears quickly (usually in a few seconds), with the clearing happening too soon to have possibly been the result of any operator action. Thus, it never met the definition of an alarm at all.
 
Chattering alarm: a fleeting alarm that repeats. This is the worst type of nuisance alarm, and is often a HUGE percentage of alarm occurrences. A popular threshold for defining chattering is three or more occurrences in a minute.
 
Now, no alarm was intended to chatter or be fleeting, and these nuisance behaviors can be fixed. Here are how alarm on-delay and off-delay times work to solve identified chattering or fleeting alarms. Most DCSs have these delay time settings, some with one, or the other, or both depending on the point type. The settings are applied to individual alarms, not “globally.” (Now, you always apply alarm deadband, if applicable, before delay time, but that is another subject! Read the Handbook for the full information.)
 
On-delay: When an alarm condition occurs, the alarm is NOT immediately shown to the operator! Instead, the on-delay timer starts (say, 10 seconds), and the alarm has to REMAIN continuously in effect for that time before it is annunciated to the operator. If it clears before the 10-second timer, it is not annunciated. So, a valid alarm only annunciates after this short delay.
 
This technique significantly reduces both fleeting and chattering alarms from being seen at all, at the price of initially delaying the appearance of a valid alarm.
 
Off-delay: When an alarm condition occurs, it is immediately annunciated to the operator. But when the condition CLEARS, that clearing is NOT immediately shown! Instead, the off-delay timer begins (say, 60 seconds), and the clearing condition is not shown unless the alarm remains clear for that interval. If the alarm re-occurs, the timer is reset. The result is that the chattering alarm becomes a single, sustained alarm with no initial delay.
 
The Handbook has far more information on the use of these, with scenarios and warnings about their proper use.
 
The problem with using these settings is in calculating them. How many seconds for each? What improvement will occur? Determining these accurately is tedious, involving looking at thousands of alarm event records for each chattering and fleeting alarm, looking at their durations and intervals, then plotting histograms. We used to supply a spreadsheet to help with this. While well worth the effort, it is kind of tedious. So, some engineers never used these settings at all, or applied some “educated guesses” to them instead of a precisely determined value.
 
PAS has long recognized this issue, and has addressed it with our Alarm Mechanic functionality. I invite you to learn more on our website. 
 
How is your organization addressing nuisance alarms? We’d love to get your comments on your nuisance alarm reduction efforts - or your favorite bad actors in sci-fi movies!