"Alarming Deja Vu"

I've never divorced and remarried my lovely wife of 29 years, but I imagine I now have something in common with those who have reconciled with their spouses after a long hiatus. I first joined the Control editorial staff in 1993 and was its editor in chief from 1997 to 2003, when I was drawn away to pursue other interests. (You can read more about those and how it will shape our coverage in the January issue.) Now I'm back after a decade at large, and it's amazing how much the process automation landscape has changed. Sort of. Like I imagine it would be coming back to your spouse.

One thing that doesn't seem to have changed much is alarm management, the subject of a recent Google+ "hangout" of sundry automation editors with representatives of Honeywell. The company's alarm management expert, Kevin Brown, said he has studies showing that our industries lose $10 billion dollars every year due to ineffective alarm management, but it's hard to get plants to give it more than passing attention.

Even the companies that buckle down don't stick to it. "End users see the problem and do a program, get results, but then slip back," Brown said. As a result, plants present their operators with as many as 4,000 alarms a day. It's no wonder that, too often, a critical problem slips by them to become a full-blown incident.

Brown asked us why alarm management doesn't get enough attention and what could be done about it, but we were just a bunch of editors "hanging out" (at least one of whom was not wearing pants), so I don't think we were much help.

I'd appreciate hearing about alarm management in your plant…if you're good at it, how do you keep it up? If not, what's stopping you? Log in and comment or if that's too public, e-mail me at pstudebaker@putman.net.

Show Comments
Hide Comments

Join the discussion

We welcome your thoughtful comments.
All comments will display your user name.

Want to participate in the discussion?

Register for free

Log in for complete access.


  • <p>I have worked on several projects in the alarm management area.<br /> Why is it a problem? It is inherently a problem of effort.<br /> Alarm management is time consuming. OK it costs $10 Billion in losses but what is the cost to best practice alarm management? What is the return on investment? Hard to quantify.</p> <p>Operators do not buy into best practices. They want to know about any problem through an alarm message not just conditions that would lead to personnel injuries or major equipment damage. Thus when a time consuming alarm analysis is done it is difficult to reduce the alarms. An alarm panel 40 years ago might have a low pressure alarm. Not we get pump stopped, low flow, and low pressure, and maybe even bad transmitter.</p> <p>Events are treated as alarms.<br /> Regulation and liability make it difficult to remove alarms. Not alarming will be considered mismanagement but almost never will too many alarms. The mentality when a single alarm is being evaluated is best to include it, the more the better. No weight is given to too many alarms, well maybe the corporate staff raises the issue of excessive alarms but usually not the operational staff. Management of change magnifies the effort to "clean-up" the alarms. The approvers often don't understand the significance or lack of to a alarm recommended for reduction of alarm status, maybe to an event.</p> <p>Once 200 loops was considered to be the maximum number an operator could handle. Now, 2000 is not unusual but with more sensors this may only be 2 or 3 units compared to 1. Operators are asked to handle a wider range of events and often manage by responding to alarms rather than monitoring the process.</p> <p>SInce the 80's plant and corporate staffs have been cut. Alarm management can be delayed until tomorrow so more urgent issues can be addressed today, It won't happen without management making it important.</p> <p>Probably all vendor's but I have worked with Yokogawa, haven't bought into alarm management even some tools have been provided. User's need to request change by providing useful suggestions to vendors. One example: Alarms such as intermittent bad input/transmitter should not considered OK until normal conditions exist for a period of time, maybe a minute. The point is vendor's need to look inside their system for conditions that aggravate alarm overload.</p> <p>Still better alarm displays are needed. While I have ideas, the vendor/user interaction could help this development.</p> <p>It remains a problem of effort.</p>


  • <p>The values of alarm limits are the biggest single factor governing alarm performance but there was never any science relating the values of alarm limits to how the process actually operated. The missing link is an Operating Envelope of the process. But an Operating Envelope (we all know what we mean but just try and draw a picture...) used to be a very difficult thing to work with so advice such as "position the alarm limits around the boundary of where you normally operate" was impractical to use. We have made it much easier with our interrogative visualisation of operating envelopes allowing limit values for many variables to be found in far less time than in the past. And because the alarms are consistent with one operating envelope there are far fewer false alarms. The new method is being used by Phillips 66 and several other companies already. There is more at www.ppcl.com.</p>


RSS feed for comments on this page | RSS feed for all comments