Alarm attributes

The first task for the measurement of alarm system performance is the determin-ation of what can be measured. Naturally, the performance of an alarm system can be most evidently captured using the attributes of an alarm message, let it be the timestamp of its occurrence, its priority, temporal length, etc. In this section, a brief overview of alarm attributes is provided, or rather, characteristics that can be assigned to an alarm message similar to [27]. Various attributes are directly assigned to alarm messages in industrial alarm systems and several more attrib-utes or characteristics can be derived from the information systems related to the operation, e.g., process layouts, process control systems, shift schedules and hu-man resource databases, etc. Table 2.1 provides an overview of the most generally accepted alarm attributes that are often present in processing industries. As can be seen in the first column, the different data sources are divided by horizontal lines in the table. However, it is important to highlight that every industrial database is unique, therefore Table 2.1 only aims to provide a general picture of information content for alarm systems and the specific definition of the source of the information or its data format is only for illustration purposes.

The first group of attributes is often stored in the Alarm & Event Log database and contains the information closely related to the message presented to or performed by the operator in the case of an alarm event. Naturally, an alarm has a tag or tag ID, a unique identifier of the alarm message. The units of the alarmed variable indicate the unit of measure of the process variable, while thealarm type or interval identifier of the alarm attribute provides information on the level of the process variable related to the assigned alarm thresholds. Besides the presence of low (L) and high (H) alarms, several other levels can be defined, in most of the cases the more severe alarms are indicated as low-low (LL) or high-high ones (HH), but these definitions are often arbitrary. In support of the operator work, a briefdescription of the problem is provided as well. The actual alarm threshold set by the operators is also present in the database. The event type describes

Source Alarm attribute Example

Event type Alarm, Return to normal,

Acknow-ledge, Operator action, Operator mes-sage, Ignore, Suppress, Unsuppress, Shelved, UnShelved

Priority Low, emergency, high, normal, journal

Suppressed Binary, 1 - 0

Consequence of deviation Pump cavitation, Tank overflow, Pres-sure drop

Consequence category Equipment, safety, environmental Corrective action Stop pump, Close inlet valve, Bypass

inlet flow Allowable response time 0.5 min, 4 min

Basis Pump cavitation at 2%, tank overflow

at 107%

Retention period 1 year, 5 years

Report requirements Pump report, Safety report, Environ-mental report Table 2.1: The different attributes and available information for an alarm

message.

what happened in the related event with the specific tag: alarmed, returned to normal, was acknowledged by an operator, an operator action was performed or the alarm was suppressed or shelved. An important indicator of a well designed alarm system is the distribution of the different alarm priorities indicating the severity of the event to the operators. Suppressing an alarm refers to any mechanism to prevent the indication of the alarm to the operator when the base alarm condition is present [14]. For example, an alarm that is caused by another alarm that is already handled can be suppressed. Shelving an alarm temporally suppresses it for a specific duration of time, and this action is usually performed by the operator.

The alarm is automatically unshelved when the duration runs out. The suppressed and shelved alarms are not raised for the operators, however, they are listed in the alarm and event summaries. The suppressing and shelving of alarm messages can be indicated by various data formats: by binary variables indicating the shelved or suppressed alarms, or by the timestamp of the start and end times or by a state indicator, which indicates whether the alarm is in an active, suppressed, shelved, or another state. In the case of alarm threshold modification, the ’from’ and ’to’

values can also be recorded.

The second group of attributes contains the original definition and implementa-tion logic of the alarm message. This informaimplementa-tion is usually stored in the alarm philosophy documents, the alarm definitions of the master alarm database and the deployment descriptions. These attributes are often not displayed to the operators during the production as they are either aware of this information or it would only unjustifiably increase their information load. As such, the originalalarm threshold of the master database is not displayed directly to the operator together with the alarm message, as only the actual threshold setting is relevant during the current situation. Moreover, in the case of a well-established and rationalized alarm mes-sage, the existence of the alarm message is well-proven, therefore, its consequence and its consequence category, corrective action, the response time for the correct-ive action and a brief explanation of thebasis of alarm message is provided in the documentation of the alarm system. Thenotification and report requirements can also be defined.

The third category of alarm message attributes describes the place of alarmed variable within the process, which is important for the monitoring of the spillover effects of malfunctions [28]. Several hierarchical decomposition structures can be defined for chemical processes, following the guidelines of the ISA-95 standard [29].

The application of the following hierarchical levels is recommended (enumerated in top-down order): enterprise, site, area, production unit, unit, and the level of sensors and actuators (although the standard focuses on the first four levels).

Therefore, if an alarm is raised, its origin can be defined hierarchically: the pro-duction units (such us columns, reactors, separators, etc.) are composed of units (furnaces, pumps,etc.). At the bottom level of the presented hierarchy, the sources of event signals are presented, i.e., the sensors and actuators.