How to perform alarm rationalization

By William L. Mostia Jr., PE

MODERN CONTROL SYSTEMS have brought us  many benefits, but along with the benefits have come problems. One of the major benefits has been the increased information available to the operator, while one of the problems is what to do with all that information.

Computer-based control systems also have increased the level of abstraction of the process. The operator has more and more information, but with a smaller and smaller window to look through, resulting in a higher and higher level of abstraction. Increased complexity and sophistication, increased automation, control concentration and separation, and additional layers of control have further increased the level of abstraction. Some systems are so abstract they approach the complexity of a video game.

To compensate for this abstraction, control systems have provided additional operator interface functions, and system designers have increased the number of alarms and alerts to help keep the operator informed. Alarms increase the amount of information going directly to the operator but they often are a source of operator overload and confusion.

In older control systems, hardwired panels were used to provide alarm annunciation. The panels were large but limited in capacity, and so by their very nature tended to limit the number of alarms. In modern control systems, alarms are generally software-driven and are essentially "free" for existing process variables. Little incentive to limit their creation has led to a laissez-faire attitude toward alarms. We can configure a new alarm at the flick of a finger, and there has been a lot of flicking going on.

Also, regulations from OSHA and the EPA as well as voluntary programs such as ISO 9000 and 14000 have led to the addition of alarms, sometimes with little consideration of the effect on alarm loads at the system level.

Some notable examples of alarms causing problems include the Three Mile Island accident in 1979, where important alarms were missed; the Texaco refinery explosion at Milford Haven in 1994, where, in the 10 minutes prior to the explosion, two operators had to respond to 275 alarms, peaking at three per second; and the recent Esso Longford gas plant explosion in Australia, where some experts concluded that operators routinely ignored alarms leading up to the explosion because, in the past, ignoring them had no negative impact.

Alarming Growth
Alarm growth is a natural outcome of the increased information load and abstraction of the modern control system. However, if alarms are not dealt with in a disciplined manner, uncontrolled alarm growth can result, which can lead to out-of-control alarm systems. If your alarm system has one or more of these characteristics, it may be out of control:

1. Many alarms during abnormal situations
2. Many alarms on during normal operation
3. High alarm loading rates (alarms per unit time, alarms per operator, alarms per event, etc.)
4. Incidents or near-incidents where operators missed key data provided (or not) by the alarm system
5. A large number of high-priority alarms
6. Alarms that are on for long periods of time
7. Alarms going off and on regularly or intermittently (chattering or transient)
8. Lost count of the number of alarms
9. Lost track of alarm setpoints or why they were set there in the first place
10. Don't know which alarms are safety, operational, environmental, informational, etc.
11. Operators don't know what particular alarms mean or that there may be inappropriate alarms
12. Operators don't know what to do when a particular alarm occurs
13. Don't know when the alarms were tested last
14. Alarms that are not useful and even confusing or obscuring
15. A large number of defeated alarms
16. No procedure or policy on alarm creation; i.e., anyone can create an alarm or change the limits on his or her own authority
17. Alarm documentation is out of date or nonexistent
18. No written procedures or policies on alarms.

Elements of Alarm Management
Driven by hazardous events, out-of-control alarm systems, and desires to optimize alarm systems, effective alarm management has moved to the front of the minds of many users and manufacturers. There can be both problems and treasures buried in the alarm system. Alarm management can be viewed as part of a larger scheme of normal operations and critical or abnormal condition management.

Alarm management is the management of alarms throughout their lifecycles. Some of the things required of alarm management are:

• An alarm philosophy
• Alarm ownership
• Controls on the creation of alarms and their functionality (from alarm activation to operator detection, operator action to perform, time in which to perform the action, etc.)
• Pre-alarm management
• Alarm change management
• Alarm prioritization
• Alarm definitions
• Alarm organization and presentation
• Alarm filtering and suppression
• Informational alert management
• Alarm operating procedures
• Training
• Alarm maintenance and testing
• Alarm rationalization for existing systems and as part of continuous improvement.

It should be noted that the control system equipment's capabilities, as well as any third-party add-ons you have or will have, can affect what you can do with an alarm system and thus can impact alarm management procedures and practices and alarm rationalization.

Rationalization Is Key