Alarm philosophy guide
ISA-18.2 and its international form IEC 62682 describe alarm management as a life cycle: you write an alarm philosophy, decide which alarms are needed, design and implement them, operate them, and keep measuring whether operators can cope. This page shows which YTCC setting or indicator belongs to each stage. The figures quoted from the standard are starting points for your own philosophy document; YTCC does not enforce them.
The life cycle in YTCC
| Stage | In YTCC |
|---|---|
| Philosophy | A document of your own: what deserves an alarm, the four priorities, response times, the targets on this page |
| Identification and rationalisation | For each alarm: consequence, operator action and time to respond, which give the priority; the alarm text; the area |
| Detailed design | Levels, Hysteresis, On delay (s), acknowledge mode, Suppressed by (signal); see Define alarms |
| Implementation | Alert classes and groups (Alarm classes and groups), alarm objects on screens (Put alarm objects on a screen) |
| Operation | Acknowledging and shelving by operators; see Shelve alarms |
| Maintenance | out of service on the alarm definition |
| Monitoring and assessment | Last 10 min, FLOOD, CHATTERING, the Alarms.* counters and the alarm history |
| Management of change | Every applied configuration change is in the configuration audit; see Diagnostics and audit |
How priorities, states and acknowledge modes behave is described in How alarms work.
Priority distribution
An alarm gets its priority from its alert class: 80 and above is Critical, 60 to 79 High, 20 to 59 Medium and below 20 Low. Choose the class during rationalisation from two questions: how severe is the consequence if the operator does nothing, and how much time does the operator have to act. The importance of the equipment alone is not a reason for a high priority.
ISA-18.2 suggests roughly 80 % low, 15 % medium and 5 % high for the alarms that operators see. With four priorities, keep Critical for a very small number of alarms, well under 1 %. If most of your alarms are High or Critical, the priority no longer tells the operator what to handle first.
Keep the configured distribution in your rationalisation list. For the alarms that actually came, download the alarm history as CSV and count the Priority
column. The counters Alarms.Summary.Critical, High, Medium and Low show the active alarms of each priority at any
moment.
Chattering and fleeting alarms
A chattering alarm comes and goes repeatedly in a short time; a common measure in ISA-18.2 practice is three or more arrivals within 60 seconds. YTCC uses the same default and marks such a row CHATTERING. A chattering alarm should be fixed in its definition, not shelved for good:
- Hysteresis stops an alarm that goes and comes again while the value hovers around the limit. It is in engineering units and applies to all levels of the signal. Common starting values are about 5 % of the range for flow and level, 2 % for pressure and 1 % for temperature.
- On delay (s) stops alarms from short excursions and from noise: the condition must hold for this time without a break. Common starting values are about 15 seconds for flow and pressure and 60 seconds for level and temperature.
- Suppressed by (signal) removes alarms that are expected in a state of the plant, for example low flow while the pump is stopped.
An alarm goes as soon as its condition clears, so use hysteresis against an alarm that goes too early.
Shelving, suppression and out of service
| Mechanism | Who | In YTCC | Rule for your philosophy |
|---|---|---|---|
| Shelving | Operator | A reason is required. The longest duration is 24 hours by default; Until manually unshelved is also offered. It is recorded in the alarm history with who and why. | Shelve only nuisance alarms whose cause is known, for the shortest time that is enough. Do not use Until manually unshelved. Review Shelved at each shift change. |
| Suppressed by design | Engineer | Suppressed by (signal) hides the alarm while another signal is not 0 | Use it for alarms that are meaningless in a plant state, such as equipment stopped |
| Out of service | Engineer | out of service on the definition; the change is in the configuration audit | Use it for maintenance, with a work order, and put the alarm back in service afterwards |
Shelved, suppressed and out-of-service alarms are still evaluated and recorded, but they are not shown, counted or sounded. Communication, command and system alarms cannot be shelved; see Communication, command and system alarms.
Alarm rate per operator
ISA-18.2 gives these targets for one operator position:
| Measure | Target |
|---|---|
| Alarms per 10 minutes, average | About 1 is very likely acceptable, about 2 is the most that is manageable |
| Alarms per day, average | About 150 is very likely acceptable, about 300 is the most that is manageable |
| Alarm flood | More than 10 alarms in 10 minutes |
| Time spent in a flood | Under about 1 % |
| Chattering alarms | None |
| Alarms standing for more than 24 hours | Fewer than 5 |
YTCC shows the rate on the runtime screen:
- Last 10 min: N at the top of the alarm list counts the alarms that came in the last 10 minutes.
- FLOOD appears next to it while more than 10 alarms came in the last 10 minutes. Each alarm that comes during a flood is marked FLOOD in its row.
- CHATTERING marks a row whose alarm came 3 times within 60 seconds.
- The counters
Alarms.Summary.*,Alarms.Areas.<area>.*andAlarms.Groups.<group>.*give the active, unacknowledged and shelved alarms for the project, each area and each group. Bind them on an overview screen; see Alarm counters.
The indicators count only signal alarms that are shown. Shelved, suppressed and out-of-service alarms, communication, command and system alarms, and alarms restored after a restart are not counted. The flood limit applies to the whole project, not to one operator. If several operators share a project, set the limit for their combined load, or watch the area counters of each operator's areas.
The limits are server settings, read by the process manager (scada-manager) from its environment when it starts and
applied to every project on the server:
| Variable | Default | Meaning |
|---|---|---|
ALARM_FLOOD_PER_10MIN | 10 | A flood is more alarms than this in 10 minutes |
ALARM_CHATTER_COUNT | 3 | Arrivals that make an alarm chattering, from 2 to 100 |
ALARM_CHATTER_WINDOW_S | 60 | The time window for the chattering count, in seconds |
ALARM_SHELVE_MAX_H | 24 | The longest shelving duration in hours, at most 744 |
For daily rates, the time spent in floods and the most frequent alarms, download the alarm history as CSV and evaluate it; see Search and export the alarm history. How far the history reaches back depends on the alarm archive; see Alarm archive and restore after a restart.
Good to know
- Alarms reach operators on the runtime screen: the alarm list, the alarm bar and the horn.
- Shelving needs the Acknowledge alarm permission.