IGNORANCE IS BLISS

NEWSLETTER ARTICLE

If taking aspirin masks the pain from a brain tumor to the point that you don't get treatment, did the aspirin help? The emphasis in alarm management on elimination of "bad actor" alarms may be a similar phenomenon. It has become a mantra that the first and biggest improvement in alarm management is to remove the nuisance alarms, ones which actuate frequently and have no operator response. Take an aspirin and call me in the morning.

Consider the case of a set of hydroprocessing units in a major refinery. Over the course of a month, the console operator averaged 4.1 alarms per hour. This was the result of a conscious effort to eliminate the "bad actor" alarms, an effort that has resulted in a reasonable level of steady state alarm activity. Everything is much better, until a minor unit upset occurs.

In this instance, it is a loss of excess hydrogen from a pentane isomerization unit to the other hydrotreaters. The resulting alarms, in five minute increments, are shown in the figure below. Over the two hours of this event, the console operator averaged 24 alarms per five minutes, or about five alarms per minute. These rates are well above most guidelines for alarm response.

This was not a major upset, more of a bump in operation. The event was handled without incident. However, how confident should this plant be of the success of their alarm management approach (i.e., the reduction in "bad actors")? How confident are you that this approach will allow your operators to handle upsets without further incident?

While this type of analysis certainly has its place in an alarm management program, it is (or should be) only one aspect of a comprehensive approach that should include alarm philosophy development; point by point reviews; documentation of alarm causes, responses and consequences; and application of conditional logic.

Copyright Beville Operator Performance Specialists, Inc.