Over time, the number of alarms in a large, highly automated petroleum refinery can balloon to 10,000 per day—so many that it can be difficult to operate efficiently. Prioritizing those notices is a big challenge for teams tasked with analyzing all those warnings. Research has shown that it’s difficult for operators to handle more than 40 per day.
When a Mediterranean oil and gas company decided its alarm system needed revamping, project managers first focused on the problem elements that prompted the largest number of alarms.
“When you start the analysis with the bad actors, in three months you can have significant results,” said an oil and gas engineer who responded to Automation World’s survey on alarms. “A project for alarm management involves a lot of different departments and you need management commitment and support to be successful.”
One technique for eliminating alarms is to prevent, suppress or manage the cascade of alarms that comes with major events like a power dip or the shutdown of a major compressor. A compressor shutdown, for example, will often be followed by warnings related to its lubrication or cooling system.
“In this situation, there must be advanced logics implemented to group alarms together or suspend some of them that are not important,” the engineer continued. “If a compressor trips, you don’t need any more alarms relevant to the compressor until it is working again. These advanced techniques are not easy to design and implement. There must be a hazard analysis to be sure that you will not miss an important alarm.”
Overall, responses to the survey indicate that, in the past five years, nearly one-third of respondents have installed new alarm management systems, while another 17 percent have reduced the number of alarms their systems use.
Re-examining alarms can bring big benefits. Four of 10 respondents said that making changes meant operators were better able to react to alarms. More than a quarter said they had fewer critical alarms, while another 27 percent had fewer standing alarms. These changes are also driving modifications to human machine interfaces (HMIs) and SCADA systems from equipment designers and integrators. They’re drawing more precise lines between true alarms that need prompt attention and alerts that don’t require any immediate action.
“In years past, alarms were created for a wide range of operator notifications,” says Kim Van Camp, product marketing manager, alarm management, at Emerson Process Management (Emerson Process Management, http://www2.emersonprocess.com/en-US/Pages/Home.aspx). “What’s new is an increasing discipline in making a clear distinction between true alarms—the notification of an abnormal equipment or process condition requiring an operator response—from all other types of operator notifications such as prompts, alerts and messages where conditions are either not abnormal or operator action is optional.”
That strategy takes human nature into account. When people get too many alarms, they often choose to ignore some of them. If critical warnings are ignored, problems will ensue.
“It is generally a poor practice to incorporate alarm escalation, in which an unaddressed alarm re-annunciates or escalates in priority,” says Bill Hollifield, principal consultant for PAS Inc. (PAS Inc, http://www.pas.com/). “Such a practice diminishes the effectiveness of the overall alarm system.”
Another benefit of keeping the number of alarms low is that when there are fewer items to learn, it’s simpler to train operators. More than half of the survey respondents said operators were only trained when new systems were launched or new hires were brought on board. More than a quarter only have annual training, and fewer than 10 percent update staffers every two to five years.
While training may not occur often, the importance of training is acknowledged by survey respondents and equipment suppliers. More than 40 percent of the survey respondents put operator training among the three most important aspects of alarm management.
“Train operators to stop ignoring standing alarms. They should report them instead. Review alarms configuration,” one survey respondent says. Another respondent says that “continuous reviews and operator training” are among the most important issues.
The need to train employees to quickly understand alerts and take appropriate action is becoming more vital as automated systems run faster and the costs of unexpected downtime rise. Equipment makers note that it’s becoming increasingly important for operators to be ready when alarms occur.
“The allowed time for the operator to take corrective action for an alarm is decreasing,” Van Camp says. “Many companies now set specific guidelines for maximum allowed response time in their alarm philosophy documents. In general, the guideline for maximum time to respond to a low-priority alarm is in the range of 30 minutes to one hour.”
The interface factor
HMIs are becoming more sophisticated, providing more information that could help operators reduce the number of failures. But as equipment gets more complex, it becomes harder for operators to remember the desired actions needed for every alarm—especially those that occur only rarely.
“A significant change in industry regarding alarms and alerts is the addition of enhanced visualization in the operator HMI,” says Hector Perez, high-performance HMI manager at PAS. “The application of better visualization enables the operator to notice small process deviations and correct them before an alarm or alert occurs.”
Going forward, operators might be required to take even more actions. For example, predictive diagnostic tasks are seeing more use as equipment designers and plant managers better understand how to use monitoring data to foresee problems. With prognostic technologies, operators can schedule repairs before alarms notify them of imminent failure. Though prognostics is a fairly new area, it’s expected to see solid growth as sensor technologies evolve, adding another layer of information that operators will need to understand.
“Prognostics can come in different forms such as operator situational awareness or continual analytics packages,” says Matthew Poplawski, product marketing manager, operator experience at Emerson Process Management. “Situational awareness is improved using overview graphics in the HMI and good use of color. Continual analytics use advanced sensors and mathematical techniques such as principal component analysis to predict when the process is moving to an unstable situation early enough for an operator to respond before an abnormal situation develops.”
Top of mind
Once alarms are built into systems, however, they can quickly become a part of the landscape that isn’t examined until something happens. The largest percentage of survey responses, 40 percent, said that they only review and/ or document alarms when something happens to demand a review.
Of those who do reviews on a regular basis, the largest segment—just over one quarter—run checks monthly. About 20 percent perform reviews daily, while 13 percent review alarms every week.
Though many managers attempt to reduce the number of alarms, there will often be instances when alarms must be added. Slightly more than half the respondents said their companies have no formal system in place to introduce new alarms. Many of those who have a plan say it’s an ad hoc process, occurring when a significant issue requires response. Others say that they usually only add alarms when equipment is changed.
“By monitoring and analyzing the history of alarms, many companies are able to identify recurrent root cause conditions and identify steps to prevent them in the future,” Van Camp says. “While not always preventable, it is often possible to minimize the impact and shorten the downtime through in- context presentation of help for the operator, including probable causes for the alarm and guidance on verifying and correcting the condition.”