December 2009

Our plant operators need HELP

By Joseph Alford

Automation and process engineers have, collectively, specified and configured millions of alarms into today's automated industrial plants. Theoretically, each alarm is intended to represent an abnormal situation requiring a response. 

Reducing nuisance alarms

An unintended consequence of the alarm implementation effort is the large number of nuisance alarms that exist in most plants, often hundreds or thousands per day. These can exist for many reasons, e.g., poor engineering design of the plant, not suppressing alarms when they are not applicable, chattering alarms, and use of embedded PLCs (within equipment) containing inflexible hard coded alarm logic.

Imagine you are a homeowner with a home security system that alarms due to the inadvertent action of a family member. Multiply this aggravation several fold, and you can imagine the frustration operators sometimes feel with the proliferation of nuisance alarms sounding horns and blinking displays on their consoles. Not surprisingly, operators have been known to lose respect for the alarm system, ignore alarms, and start missing legitimate abnormal situation alerts.  

If anyone doubts the ability to improve alarm systems, consider the complex automobile containing thousands of components and many electrical-mechanical systems. It is hard to think of more than about five configured alarms for automobile drivers (e.g., low gas level, seat belt not fastened). None are redundant, and rarely is an alarm generated when not appropriate.

Let us help our plant operators by pursuing continuous improvement projects, focusing on eliminating nuisance alarms. In most cases, it is only a small percentage of configured alarms that account for the majority of a plant's total alarms. And, by the way, addressing the root causes of nuisance alarms just might improve plant reliability and consistency.

Alarms that "guide"

Alarm management literature notes the three main attributes of an effective alarm are to 1) alert, 2) inform, and 3) guide. With some exceptions, alarm functionality in most plant automation systems does a good job of "alerting," a fair job of "informing," and a poor job of "guiding."

The "guide" attribute of an effective alarm is based on each alarm requiring a response. What is an operator to do when there can be multiple root causes of an alarm and the response is different for each root cause? Given the large number of configured alarms and associated root causes, the response landscape is often more than operators can be expected to memorize or learn in training classes.

Utilizing hard-copy procedures is not satisfactory either, as it takes too much time to locate and read procedures in times of plant crises.

Let us help our plant operators by prospectively determining possible root causes for a configured alarm (when rationalizing the alarm) and then presenting in the human-machine interface (HMI) an alarm's expected response, or sequence of responses in order of descending likelihood, in cases where multiple root causes and responses can exist.

Other alarm system needs

Operators are further frustrated with information overload in times of plant crises. Many alarms are redundant with others, are not prioritized, and are mixed in with informational messages. Some systems do not permit programmatically changing alarm attributes as a function of process step/phase (e.g., priority), which are often needed for batch. Some systems are not effective at alerting intended alarm recipients located in remote locations.

Let us further help our plant operators by providing them with an HMI consistent with alarming best practices, including properly prioritizing alarms, minimizing redundant alarms, and clearly differentiating alarms from informational messages.

More capable and efficient operators

Finally, has anyone noticed plant automation and engineering support staffs are being reduced in many companies? This can only result in greater dependency on operators to do low-mid level data analysis and process troubleshooting. At the same time, a large number of experienced operators are now retiring-carrying their process knowledge with them.

Let us help our plant operators regarding these paradigm shifts by capturing the knowledge of our experienced operators and complying with the recently approved ISA/ANSI-18.02 standard (Alarm Management for the Process Industries).


Joseph Alford, Ph.D., P.E., CAP ( retired in 2006 after a 35-year career in process automation at Eli Lilly and Co. He is a Fellow of ISA and AIChE, a recipient of ISA's Douglas Annin Award, and a 16 year member of the InTech Editorial Advisory Board.