ALARM SYSTEM AVALANCHE AND MULTIPLE PROCESS CONTROL

NEWSLETTER ARTICLE

Today, it is fairly common for operators to control multiple processes. In some cases, the new computer control systems have increased the span of control of the operator five-fold. The downfall of many of the computer control systems is the avalanche of alarms that occur during any type of transient, or upset situation. The alarm system at Three Mile Island was a major contributor to the severity of the accident that occurred there. A transient situation in a nuclear plant will produce 200-300 alarms in the first couple of minutes.

The result of an alarm avalanche is that the operator's information processing capacity becomes overloaded ("A Procedure for Reviewing and Improving Power Plant Alarm Systems", EPRI 1984). With many of the new process control systems the alarms are printed out sequentially by the time of activation on an alarm summary display, such that the alarms scroll down as new alarms actuate. Many operators have discovered that during upsets, the alarms scroll down at such a rate that they can not possibly read the new alarms, let alone interpret their importance to the process.

A consequence of alarm avalanche is that an upset on one process unit could cause an upset on another, not because of the interaction of the processes, but rather because the processes have the same control operator. What often happens is that the operator is engrossed in one process, and misses a situation developing on another process. Many managers have expressed concern over this upset propagation that can occur.

A solution to the alarm avalanche phenomena is to prioritize the alarm system. Alarm prioritization allows the operator to ignore many of the nuisance alarms that occur during upsets, and focus their attention on the critical and high priority alarms. Beville has developed and implemented alarm prioritization schemes at various processing plants. Several steps that must be taken to prioritize an alarm system include to first define the alarm group classifications, such as what is meant by a high or critical priority alarm. The second step is to identify how the alarm priorities will be configured into the alarm system (i.e., which priorities go into which groups). The next step is to assign the priorities to the alarms and finally configure the alarms into the system. As the priority system is developed an alarm response manual should be developed to capture the logic used, as well as the alarm meaning and typical operator responses to all of the alarms.

Copyright 1989 Beville Operator Performance Specialists, Inc., All Rights Reserved