Is Annunciator Evolution Overdue?
Though they’re often called Christmas trees, most annunciators don’t spit out presents when they light up.
Most simply let control room operators know something is wrong. However, annunciators that only announce problems aren’t enough for users of today’s increasingly complex and integrated automation systems. Fast, coordinated information is crucial to help operators correctly handle many current process problems, and so some annunciators are evolving to meet their demands.
So far, annunciator evolution has consisted mostly of prioritizing the signals they put out. Advanced models can often log times and sequences of alarm-outs.
Choosing a proper annunciator system depends on correctly evaluating many interacting process variables. It’s also vital to assess the potential danger a process or machine may present to personnel, equipment, and materials consumed. As a result, control engineers must analyze each process segment with an eye toward safe operation. Next, they should put all the parts together again to help them predict, engineer out, and prevent dangerous situations.
When setting up an annunciator, engineers must also decide at which alarm states the annunciators will react. There are five alarm states normally cited in ISA (Research Triangle Park, N.C.) and other standards. These include most-critical to non-critical alarms, status information, and analysis of data that might trip the unit.
Seeking a crystal ball
Though they can prioritize, many annunciators still aren’t able to distinguish between nuisance alarms and truly important ones. Critical alarms may require immediate operator action, but they often can’t arrange alarm data into a format to help operators correct problems faster and more effectively.
Even today’s annunciators often can’t provide more than a log of alarm-trip times and sequences. This log usually isn’t enough to enable the operator to determine cause, effect, and cure. If properly formatted, trip analysis methods could provide some of the missing data, but it would only be available after the fact, while disaster prevention requires foresight.
Though clairvoyant annunciators aren’t available yet, new pattern recognition methods are in use in some automated vision systems and other advanced controls. These methods can help operators find formerly invisible trends, and take preventive action.
Many systems achieve pattern recognition in real-world industrial applications by using recently developed neural network technology. It may soon be practical to make alarm-out sequences work with pattern recognition statistical analysis software. The resulting analyses could help control engineers determine whether specific alarm-out patterns are significant, critical, or merely nuisances.
Once a set of desired conditions is identified, it should be simple to store applicable patterns in an annunciator. Using pattern recognition to analyze alarm-out trends should then make it possible for annunciators to predict future alarm-outs, lock out nuisance alarms, and illuminate message windows only when genuine problems occur. This would fulfill annunciators’ original mission and give operators and their Christmas trees a few more safe and silent nights.