Management comes first for alarm management success

Petrochemical plant managers have a difficult task: how do they maintain safety within the plant while optimizing plant operations, maintaining profitability and remaining environmentally friendly? One answer is a reliable alarm management system.

The essence of alarm management is to supply operators with enough information to prevent abnormal situations, and to prevent abnormal situations from escalating into unpreventable situations.

Research from Honeywell estimates the inability to diagnose and control abnormal alarms costs manufacturers $10 billion annually in the U.S. petrochemical industry alone. These losses are caused by accidents, equipment damage, unplanned plant or unit downtime, off-spec production, regulatory fines and intangible costs related to environmental and safety infractions.

Decades ago, hard-wired alarm systems alerted operators to problems. Adding more alarms into the system meant running more wires at a higher cost. New alarms were gradually incorporated into the existing systems. Operators were occasionally flooded with alarms and diagnosed abnormal alarms by using strip charts, panel lights and field operator support.

Photograph courtesy of Matrikon, Inc.

Identifying minor alarms became another major problem. An operator is efficient only when he understands the meaning of an alarm and how the consequences of not responding to one can affect the plant floor. For the most part, engineers are unaware what the consequences might be if alarms are ignored.

To help operators assess alarm situations properly, they need a online documentation system that alerts them on the severity of alarms across the plant floor. Displaying the amount of time that can elapse without responding to an alarm before it escalates to an urgent level is critical. Operators also face lack of experience in alarm management.

Photograph courtesy of Matrikon, Inc.

In “Institutionalizing Alarm Management,” Gould indicated that not knowing how to establish and execute an alarm management project leads inexperienced operators to put off alarm management activities in favor of 'higher priority’ issues. Identifying problems with an alarm management system is easy to do. The difficulty comes when the operator needs to identify the necessary solutions to stop the alarm and fix the problem.

When configuring alarm systems, it is important to keep in mind the design used in the operator’s process interface. In past events, refineries suffered from major catastrophes due to the design of the display systems. These systems were not automated. The designs were simple interfaces that, in most cases, were configured improperly, adding confusion to the inexperienced operator using the interface. Configuring the interface properly eliminates problems derived from alarm flooding and wasted CRTs (see graphic, page 56). Proper alarm and display system design is obtained by determining the content to be displayed first, and then designing the form and look of the alarm.

Figure 1

Plant floors need alarm management systems to ensure that the processes within the plant are running safely and efficiently. However, if the system is poorly configured and runs inadequately, the same system that is essential for the safety of the plant could be the main source of the plant’s alarm problems. According to Beville Engineering, some of the characteristics of poorly configured alarm systems include:

• Alarms that have no operator action
• Alarm conditions that have multiple alarms
• Alarms that are 'in’ during normal operations – violation of the 'dark screen’ principle
• Alarm actuation rates in excess of the operator’s information processing capabilities
• Alarms with improper setpoints or deadband, causing alarm cycling
• Excessively high alarm-to-controller ratios
• Alarm priority distribution that has too many emergency and high-priority alarms and too few low-priority alarms.

The 'dark screen’ principle is important for obtaining an alarm system that alerts the operator only when something is wrong. When designing an alarm system, this should be the main principle to follow. Ideally, no alarms should be presented when a process part is in a normal state and without failures.

However, during normal changes of the process, a number of parameters are changing, and as long as the variations are normal with respect to the state of the process, no alarms should be presented. The main goal is to avoid information overload and unnecessary distractions in all states of the process.

The goal of alarm management is to reduce the number of alarms requiring operator action. By implementing alarm management tools, operators and other plant personnel are sure alarms are always consistent with fundamental plant constraints. At the same time, operators and other plant personnel can efficiently send and receive notification reminders without cluttering the alarm system. Event analysis tools help identify these cause-and-effect relationships. Such tools analyze alarm and event activity metrics, differentiating alarm system problems and performance and identifying operator workload.

In an ideal world, each alarm would prompt a distinctive operator action without having several alarms triggered by the same action. Alarm response analyses configure alarms to actuate only when necessary. Alarm response analysis systematically gathers information on each alarm and uses this data to establish the points that need to be addressed, eliminated or modified. Cataloguing each point provides the necessary documentation for changes and regulates future alarm add-ons. Proper alarm management configuration demonstrates the plant’s commitment to its safety and environmental well being.