How to perform alarm rationalization

Table I: Top 10 Reasons to Rationalize

Rationalization also can, in some cases, identify the need for new alarms or changes in the process, equipment, or instrumentation. It can be done to fine-tune an existing good alarm system, but it is more commonly done where the alarm system has gotten out of control.

Note that alarm rationalization is not a one-shot process. The forces of chaos are out there looking for any opportunity to take control of any complex system, and alarm systems are no exception. Over time, people will come and go, the process will change, operating philosophies will change, marketing will stick its nose in things, the hardware system will change, improve, degrade, etc. All these are opportunities for changes or lack of changes to the alarm system, and they indicate the need for periodic alarm rationalization. Training, procedures, procedural controls, and auditing are some of the tools used to maintain an optimum alarm system for effective and safe operation of a plant.

How to Rationalize Alarms
Alarm rationalization is a structured process that generally involves an approach similar to that of a HAZOP team, with representatives from operations, maintenance, engineering, and safety. It is important to have operator input on this team. It is also important to have an organized plan to perform the alarm rationalization, with an established procedure and practices.

While alarm rationalization will vary from company to company and plant to plant, the methodology generally consists of eight basic steps (Figure 1). These steps are presented serially but in fact can overlap or run in parallel in some cases.

Figure 1: An Eight-Step Program

The steps on the path to alarm rationalization are shown in series, but in some cases may overlap or be performed in parallel.

1.  Develop an Alarm Management Procedure
A consistent, comprehensive alarm management procedure/philosophy is necessary before beginning alarm rationalization.

The procedure typically contains a plant alarm philosophy, alarm type identification methodology (operational, safety, environmental, etc.), risk identification methodology and method of prioritization of alarms, alarm functionality requirements (presentation, organization, pre-alarms, operational time requirements, etc.), alarm filtering or suppression methods, identification of undesirable alarm types and how to handle them, methods of setpoint determination, testing requirements, alarm sequences, acceptable alarm metrics, documentation requirements, etc.

While alarm rationalization is only a part of the overall alarm management, it is absolutely necessary to have defined consistent alarm practices to apply to the alarm database.

2.  Develop Alarm System Metrics
It is hard to measure your progress if you don't know where you start and where you end up. To measure the progress of alarm rationalization, you need to develop alarm system metrics for your system. Typical metrics can be total number of alarms, alarms per operator, alarms per hour, alarms per identified abnormal situation, fraction of unacknowledged alarms, average time for an alarm to return to normal, average number of active acknowledged alarms, number of chattering alarms, number of standing alarms, number of nuisance alarms, number of disabled or shelved alarms, etc.

Operational metrics also can be used, such as total production rate, off-spec production, number of upsets, uptime, workload, etc.

Some safety and environmental metrics that could be used are the number of plant shutdowns, number of incidents/near misses, releases to the atmosphere, releases to flare, etc.

Operators and operating staff interviews also should be used as metrics. These people are on the front line and have good firsthand information on how the plant is operating.

3.  Benchmark the Existing Alarm System
This is a two-part process. First is system benchmarking, which consists of an alarm operational analysis (statistical analysis of the existing system) and identification of the current values or states of other alarm rationalization benchmarks.

Statistical analysis of the current alarm activity of the existing system should look at alarm frequency, duration, occurrence, bursts (floods), correlation (or not) with process or equipment activity, correlation with other alarms, acknowledgement time, etc. This type of analysis can be a static analysis of the alarm and operational logs, a dynamic online analysis via resident software that monitors alarm activity in real time, or both.

Static analysis typically has the advantages of working on all systems; working with existing prior history to determine metrics; being flexible in the metrics used; correlating with past or current operational, process, or equipment events; and using a human to review the data who may (or may not) catch trends or issues that would escape an automatic system.

Disadvantages are the analysis is typically done by hand and can be cumbersome, and some trends may be missed.

Dynamic analysis advantages typically include statistical alarm activity detected online with built-in alarm metrics calculated both in snapshot and over time, built-in alarm-system expertise that can complement the user's capabilities, automatic documentation to meet regulatory requirements such as 21 CFR Part 11, and management of alarm parameters. Dynamic analysis can be used to monitor the system after rationalization to detect downstream issues, provide alarm and process event analysis, and, in some cases, analyze manually input alarm data.