INDUSTRIAL MACHINES are performing ever-more complex operations. HMIs can provide an enormous amount of operating and predictive data to operators and beyond. And those operators often monitor entire banks of machines, one hopes, mostly just making sure things are going right, though they still need immediate, accurate advice when things go wrong.
When minor problems happen, the price to pay, while rarely insignificant, usually is bearable. When major things go wrong, operator and asset safety can be at risk. Translation: your alarm system better be well-configured and accurate.
Earlier this year, we received a copy of The Alarm Management Handbook, coauthored by Eddie Habibi and Bill Hollifield of PAS, a company involved in advanced process control and condition management. The book is aimed at process industry end users who might deal with hundreds of alarms each day, but it has some nuggets that machine control designers, who have moderate to highly complex operating processes, will find helpful when they consider the alarm structure of their systems.
The 168-page book covers how to justify alarm management, deciding what should be an alarm, real-time alarm handling methods, and the future of alarm management.
Most importantly, the book reminds us that no amount of alarm management will replace the constant surveillance of a qualified operator, and that operators must be trained on the alarm management strategies to enhance their ability to make judgments based on experience and skill.
Does an event require operator action? ask the authors. They advise readers to allow alarms to be generated only for events that require operators action to prevent or mitigate upsets and disturbances. If thats not the case, they say dont light up displays with unnecessary information.
They add that alarm solutions from different suppliers might have differing nomenclature for priority levels. The best practice is to use three levels of annunciated alarm priority, add Habibi and Hollifield. Humans are wonderfully able to understand things that are in three categories. Four or five are the max, they add, and anything beyond that becomes cognitively blurred and confusing, i.e., not helpful. Multiple alarms never should signify the same thing.
In addition, the authors urge users to create an alarm philosophy. It covers both new systems and modifications to existing systems, they state. Its for both in-house use and contractor use during projects. It also should meld with your customers schemes as well as possible.
The book also asks readers to think about the graphics display for their control system. Principle #1 is that the graphics should be boring. This runs counter to the ability of software to provide colorful, animated renditions of moving conveyors, rotating shafts, etc., often with inconsistent color coding and navigation from screen to screen. Avoid the kinds of graphics that look impressive when kids visit on career day, or when the corporate vice president comes by on a tour, caution the authors. The operator who uses them every day will be using a very poor set of tools to accomplish the primary job of operating safely and profitably. They add the graphical elements and controls must behave and function consistently.
Their advice is clear: good graphics have gray backgrounds to minimize glare. Dont use animation. Make limited use of color, and use alarm colors to display alarms and nothing else.
For a multiple-machine operator panel, provide a layout that is generally consistent with an operators mental model of the process, say the authors. The operators might relate better to the physical plant layout than the layout shown on a P&ID.
The authors also stress the importance of providing trend information in the operator display. The best operators actually run from trends, they add.
