Make the Most of Alarms

As processes have become more complex and dynamic, alarm management software has evolved, and now provides deep intelligence to keep plant personnel safer, more aware and more efficient

There can be little doubt that effective alarms are one of the most critical elements of safe, high-efficiency operations. Alarms not only keep people safe by alerting operators when a process has moved outside of its defined boundaries, they are also a critical check on the steps of an operation to ensure consistent manufacture of only the highest-quality product. While designing, building and implementing an effective alarm strategy has never been easy, the complexity of this effort increased dramatically with the introduction of digital control systems.

One of the key benefits of modern digital control systems is flexibility, and that flexibility extends to the creation of alarms. Whenever anything goes wrong with operations, a user can quickly step in and create an alarm to ensure they have more visibility to that problem in the future. However, the ease with which users can add alarms often leads to an overwhelmingly complex array of different alarms, which in turn complicates and confuses operators both as they run the plant, and when they try to manage the alarms in the system. Users can often create, propagate and change alarms faster than documentation can keep up.

The ability to propagate alarms faster than they are documented is a serious hinderance in today’s world of complex, multi-state processes. In very little time, an operations team can make changes on top of changes to the alarm database, and when there is a process upset or a change of process state, operators are flooded with thousands of alarms in seconds — far more than any human can handle. Not only are alarm floods stressful to the operator, they also mask real problems in a torrent of non-issue alarms.

Implementing alarm management is the key to bringing alarms under control. However, not all alarm management solutions are created equal. Manual alarm management is cumbersome and complex, and often leads to stalled projects. Alarm management software can help, but only if it is designed for working with modern control and supervisory control and data acquisition (SCADA) systems.

Fortunately, today’s best-in-class alarm management software solutions are designed to dramatically reduce the time and effort of configuring and maintaining alarm management. As those solutions evolve with emerging tools like artificial intelligence (AI), they will continue to improve in the coming years, eventually becoming autonomous copilots that help operators manage their processes far more safely and efficiently (Figure 1).

Alarm management failures

Traditional alarm rationalization — without the aid of software — required engineers to export all the control system configuration into files. The team would then need to manipulate those data to enable the data to be moved to a spreadsheet. Once the data set was in a spreadsheet, it would need to be organized to ensure it was in a usable format. All these steps preceded the actual rationalization.

After the spreadsheets were created and organized, teams of engineers would typically spend months trying to rationalize the alarms and adjust the spreadsheets accordingly. On average, this process required the team to make 1,000 to 2,000 changes to bring alarms in line for a typical process plant. Moreover, as each of these changes was made, any documentation needed to be performed manually for each instance — an extremely time-consuming process that was often ignored or overlooked.

Today, many of the steps of rationalization are performed using a software solution, which makes it easier to eliminate mistakes. However, not all software solutions are created equal. Many alarm management solutions require users to perform all rationalization offline, and then apply changes manually. Doing so is time consuming, complex, prone to error and makes change management difficult to implement and follow. As a result, most current alarm management projects fail not because the team did not complete rationalization, but because once they finish, they have so many changes to manually enter into the control system that they never complete the job.

First and foremost, it is difficult to decide how to get the changes back into the control system and who will do it. Often, weeks or months go by with some changes implemented and others not, and the team never truly knows what has changed. In the meantime, processes change, and new alarm edits are made on top of previous changes. Ultimately, the team can never catch up.

Even if the team does successfully complete rationalization with an offline software tool, maintaining those changes becomes a herculean task. In the coming months or years, the plant will likely have equipment changeouts, or necessary process changes.

If all alarm rationalization is performed offline, it becomes increasingly difficult to keep up as more and more changes are applied. Each change must be recorded offline, and then applied manually to the control system. Moreover, if a year later the team wants to know if the alarm system is still valid, they need to go through the process of exporting data and comparing again — a process that will likely take weeks or months.

 

Integrated software

To avoid the complexity of transferring offline changes to the control system manually, many of today’s forward-thinking organizations are leveraging alarm management software that is integrated with their control system by design. Seamlessly integrated alarm-management software dramatically reduces the time engineers spend updating the control system with new alarm parameters. Such solutions empower teams to perform rationalization online while connected to the control system.

With an integrated alarm management solution, alarms can be modified one by one, or in bulk. As each change is made, the user can see the control system configuration and establish process boundaries linked directly to assets, and then easily record that change in the documentation. If an asset or alarm changes at a later date, the team can see what assets were impacted by that change, and it can audit the change against the boundaries.

Most importantly, once the team has made its changes in an integrated alarm management solution, they simply click a button and all changes are applied to the control system. The ability to apply changes online saves significant time and effort. For example, a large integrated chemical site can make about 30,000 parameter changes per month to their control system, many of which impact or are related to alarms. It would be nearly impossible for that process manufacturer to keep up with those changes without an integrated software solution. But with the right software, the team can instantly see the last-read value when a change was made, what was approved, who approved it and why, and the initial reason the alarm was created. This type of database allows the team to document and audit those control system changes quickly and easily.

 

Changes increase complexity

Historically, when engineers would design alarms for a control system, they would ask themselves three key questions:

• What alarm does the plant need?

• What is the purpose of this alarm?

• What priority should be given to the alarm?

However, even after asking those questions, many struggle with alarm floods during transition periods like startups, upsets and shutdowns. Transitional alarm floods contribute to incidents, and this issue is exacerbated because incidents are more likely to occur during transitions than in periods of normal operation. With just one additional question, it becomes significantly easier to eliminate these alarm floods:

• When does the plant need this alarm?

Modern alarm management must accommodate dynamic process states. For example, consider a flow alarm on a heater. When the heater is operational, low pass flow through a tube would be a serious concern. Over time, the tube with the low pass flow would warp and eventually create a loss of process containment, and material from the process would pour directly into the heater. Hypothetically, the process could be shut down for weeks and cost the organization millions of dollars. As a result, when the heater is operational, it is essential to receive low pass flow alarms.

However, sometimes the heater will be shut down, but if the process unit does not have dynamic alarm management, the alarm system will not know the difference. If the heater has 16 pass flows, the operator will receive 16 alarms, even though this is the expected state during shutdown. Those alarms will fill up a page and push many alarms onto another page. The low pass flow may be a high priority alarm, but it is meaningless in the shutdown operating mode.

While during shutdown, such a problem might be frustrating, during startup it becomes dangerous. During startup, operators are putting mass and energy into the process and hoping that this is occurring within specification. However, the low pass flow alarm that was activated while the heater was shut down (potentially several weeks ago) is still a standing alarm and is now valid because the operator has moved the process into an operational state. As far as the control system is concerned, it has notified the operator, even though that notification came a long time ago and may be on page 20 of the alarm screen. The system does not know the alarm was not valid before, so it will not re-alert the operator. Operators are required to maintain an incredibly high level of vigilance to track such aberrations across extended time periods, and mistakes are far more likely to occur.

In fact, most failures occur on startup because operators are effectively working without alarms. The alarm is on their screen and they are expected to know about it, but often that means scanning hundreds of alarms across multiple pages to identify if it is still valid. Doing so is not impossible, but it is a recipe for disaster.

 

Dynamic alarming

Dynamic alarm management helps eliminate the alarms that occur as a result of changing operational states. Engineers simply design the state logic — for example, run, upset, shutdown and startup — and categorize each alarm by one or more of those states. The transition technology in the most advanced dynamic alarm software enables the operators to seamlessly move between states without creating custom logic for each alarm (Figure 2).

In the heater example above, the pass flow alarm would be identified by the alarm system as not applicable during the shutdown state, so the operator would receive zero alarms instead of 16. Then, when the team transitioned to the startup state, if pass flow was blocked, the system would recognize that in the current state the alarm was valid, and it would be delivered to the operator at the appropriate time. This technology would improve visibility and increase the safety of the operation, regardless of how many times the process state changed.

 

The future of alarming is AI

Alarm management is not a set-and-forget process. No matter where they are in their alarm management journey, organizations should always be updating and improving the performance of their alarm strategy. However, in an era of worker shortages and increased efficiency and sustainability goals, finding the time and personnel to continuously manage alarms can be difficult.

Today’s most effective alarm-management-software solutions help teams manage the complexity of alarms. Operators using these tools log in and can immediately see the top ten alarms in a report, and they can then click through those alarms to receive recommendations of ways to intervene and improve the operational state. Moreover, these same tools can make basic recommendations to improve the configuration of the alarm software itself. Teams can receive notifications of chattering alarms and suggestions of alarm bands they should change to reduce that chattering. Advanced software can also help teams recognize and remedy redundant alarms to help avoid unnecessary alarms (Figure 3).

In the near future, these same solutions will begin to add more autonomous operation, with AI engines working alongside operators to help identify which alarms are useful, and which are less so, and to recommend changes to engineers to streamline the alarm database. These tools will further shorten the time required for alarm rationalization, with AI copilots that continuously identify opportunities for improvement by recommending alarm settings and changes to boundary levels. Engineers can review and approve these suggestions, spending two minutes instead of twenty to create or improve each alarm.

Moreover, AI tools will also help teams ensure documentation is as thorough, accurate and standardized as possible. Much like today’s text message and word processing tools, AI can be applied to evaluate documentation as it is being written, recommending additions or formatting to help ensure all documentation is easy to read and reference.

 

Building for the future, today

Alarm management is a safety issue. Often, the fallout from poor alarming is not obvious until a disaster strikes, and as processes get more complex, the ability to manage alarms so they keep operators alert and do not cause confusion will only become more critical. It is so critical, in fact, that countries around the world are beginning to legislate alarm standards.

Not only do the foundational technologies exist to dramatically improve alarm management, they are also typically easy to install and use. Plants that invest today in a foundation of integrated, dynamic alarm management software will not only reap significant safety and operational efficiency benefits today, they will also position themselves to take advantage of the emerging AI technologies that will soon redefine and improve how plants operate. ■

Edited by Dorothy Lozowski

 

Acknowledgement

All figures courtesy of Emerson

 

Author

Dustin Beebe is the Vice President of Performance Software at Emerson (Baton Rouge, La.; Email: [email protected]). He is responsible for the alignment of the Control Performance, Operator Performance, and Simulation businesses globally and the strategy synergy between Emerson and AspenTech. Prior to joining Emerson, Beebe served as the President of ProSys until it was acquired by Emerson in 2018. He has been in the industrial automation business since 1996. Beebe holds a B.S.Ch.E degree from the University of Arkansas in Fayetteville, Arkansas.