Hi everyone,

I'm researching alarm management in petrochemical/process plants for an engineering project.

In your experience:

• What's the biggest problem with current alarm systems?
• Is alarm flooding or alarm fatigue still an issue?
• If you could change one feature, what would it be?

Looking for practical experiences from industry. Thanks!

 

The biggest problem with current alarm systems? Cryptic alarm text messages with little or no help to decipher the alarms or conditions. Consider the alarm L.O. PRESSURE LOW for a large rotating machine. When is the alarm annunciated (at what decreasing pressure)? When does the machine trip (at what further decreasing pressure)? What has the L.O. pressure been in the last few hours before the low pressure alarm was annunciated--quickly and graphically? What can be done to avoid a trip (not EVERY possible trip, but, in general--decreasing load/speed if possible, starting a standby pump if possible; is the standby pump started when the low pressure alarm is annunciated and is it running now?). Many alarms are similarly cryptic: COOLING WATER PRESSURE LOW. JOCKEY PUMP RUNNING (my least favorite alarm--is it running when it should be running, or has it not been running when it should have been running?); why is the jockey pump running; why is this an alarm if the jockey pump starts to maintain pressure in the fire protection system and then stops when pressure is "normal" and starts again if the pressure drops? This common for conditions that are NOT really alarms--usin the alarm system to alert operators to starting/stopping of equipment that is expected to periodically start or stop.

Alarm flooding/fatique is almost always a problem--and it desensitizes operators to more serious alarms or checking alarm text messages (which may or may not be cryptic). If an alarm is "dithering"--going in and out of alarm possibly because of no hysteresis or deadband on the parameter being monitored--that is a problem which the control technicians or mechanical department should be working on to quickly and properly resolve. Alarms which might be expected to occur after some major event should be blocked (such as low hydraulic pressure after the hydraulic pump has been stopped or the hydraulic pressure relieved as the result of an emergency).

Finally, if I could change one feature, what would it be? That's a toss-up for me between the two issues above. But, operating (which includes alarm management!!!) means having access to or knowing when an alarm will be annunciated (when the L.O. pressure drops below 16 psig, for example) and if the machine/equipment will trip if the pressure continues to drop--and at what decreasing pressure. AND having quick access to recent historical data for the alarm condition (what has the L.O. pressure been for the last two, four, six, 12 hours; how fast has the pressure been decreasing from normal; etc.).