Skip to content
Thermite SCADA
Alarming & Notifications

Configure alarm conditions and route notifications.

Thermite SCADA evaluates numeric limits and boolean conditions, tracks acknowledgment and shelving, and writes alarm transitions to a JSON-lines journal. Notification pipelines route matching alarms through schedules, rosters, and escalation steps.

Alarm types

Conditions, source quality, and project scope.

Numeric limits

High, high-high, low, and low-low thresholds with engineering-unit deadbands.

Boolean state

Raise an alarm when a boolean tag matches its configured alarm state.

Quality suppression

Unreliable source quality suppresses existing process alarms until valid data returns.

System alarms

Gateway infrastructure conditions feed system alarms through the alarm engine.

Project scope

Production tag alarm definitions retain project identity for display, operator actions, and notification routing.

Policy

Controls for the alarm list.

Use priorities, hysteresis deadbands, on/off delays, acknowledgment, and bounded shelving to manage alarm behavior.

Priorities

Critical, high, medium, low, and diagnostic priorities support alarm filtering and notification routing.

Delays

On-delay and off-delay require conditions to persist before activation or return to normal.

Deadbands

Engineering-unit hysteresis controls when a numeric alarm clears after crossing a limit.

Shelving

Shelve an individual alarm for a bounded duration, with actor attribution and automatic unshelving.

Acknowledgment

Acknowledge an individual alarm. With authentication on, project roles govern acknowledgment and shelving.

Notifications

Pipelines, schedules, rosters, and delivery channels.

Pipelines

Available

Filter alarms by project, priority, area, or source-path glob and route matching occurrences through ordered steps.

Escalation

Available

Advance through ordered notification steps when unacknowledged alarms outlast configured timeouts.

Schedules

Available

Weekday and time windows select on-call recipients, with a configured default member.

Rosters

Available

Named on-call lists hold recipient contact details for the configured delivery channels.

Email

Available

Deliver alarm notifications through a configured SMTP server.

SMS & voice

Available

Deliver SMS messages and spoken alarm summaries through a configured Twilio account.

Webhooks

Available

Send alarm-notification HTTP callbacks with configured retries and optional HMAC signing.

Alarms as files

Alarm definitions are reviewable.

Alarm settings live with their tag definitions in TOML. Review limit and priority changes in Git, publish the project, then restart the gateway to apply changed alarm definitions. Notification routing is configured separately in notifications/*.toml.

# projects/press-line-a/tags/hydraulics.toml
[[tags]]
name = "HeadPressure"
path = "hydraulics/head_pressure"
data_type = "f32"
device = "press-plc-01" # example device and NodeId
address = "ns=2;s=Hydraulics.HeadPressure"
eng_units = "psi"

[tags.alarm]
priority = "high"
on_delay_ms = 2000

[tags.alarm.levels]
deadband = 25.0

[tags.alarm.levels.hi]
limit = 2200.0
When an alarm trips

Have a script do something about it.

Every alarm transition is written to a journal file as it happens, and history is one HTTP request away. That is enough for a script beside the gateway to act the moment an alarm goes active. We asked an agent for one. It wrote the script, waited for a real alarm, and this is what the alarm produced.

Agent session

You When HF-1’s high-temperature alarm goes active, save a snapshot by itself: the fifteen minutes of bath temperature and level before the alarm, as a CSV with a short summary beside it. One script beside the gateway. No changes to the project, and no restart.

Agent The gateway already writes an alarm journal with each activation’s timestamp and trip value. I’ll watch that file and read the history over HTTP.

Read site/logs/alarms.jsonl one JSON line per transition: alarm_id, transition, value, timestamp

Fetch /p/cast-house/api/history/hf1/bath_temp?start=2026-11-12T18:00:00Z&end=2026-11-12T18:15:00Z&mode=raw raw samples, each with its quality and timestamp

Write scripts/alarm_snapshots.py 158 lines, Python standard library only

Run python3 -u scripts/alarm_snapshots.py Watching site/logs/alarms.jsonl for hf1/bath_temp:hi

Agent The watcher is running. HF-1 is at 1357.2 °F and rising. I’m waiting for a real activation.

Read reports/alarm-snapshots/watcher.log Activation 2026-11-12T18:22:39.047191Z, occurrence 3 · Saved the CSV and summary: 674 samples

Agent The report pair is saved, and I checked all 674 CSV samples against the historian. Trip value 1405.4 °F, peak in the window 1410.3 °F, level at the alarm 76.2 %.

Illustrative walkthrough of a session we ran, shortened. The furnace is a simulator; the alarm, the script and its output are from the run.
reports/alarm-snapshots/hf1-bath-temp-hi_2026-11-12T18-22-39.047191Z.txt
HF-1 high-temperature alarm: 2026-11-12T18:22:39.047191Z (UTC)
Occurrence: 3; project: 01a1d088-16b1-75fe-b6ce-cb941f742be3
Window: 2026-11-12T18:07:39.047191+00:00 through 2026-11-12T18:22:39.047191+00:00 (inclusive)
Trip value: 1405.4 degF (alarm activation event)
Peak recorded temperature: 1410.3 degF at 2026-11-12T18:09:45.148Z
Level at alarm: 76.2 %; last good sample 2026-11-12T18:22:39.047Z (0.000 s before alarm)
History is recorded on change with deadbands; no interpolation. Summary uses good samples only; CSV retains every recorded quality.
Samples: hf1/bath_temp: 338; first 2026-11-12T18:07:40.148Z; last 2026-11-12T18:22:39.047Z; hf1/level: 336; first 2026-11-12T18:07:39.147Z; last 2026-11-12T18:22:39.047Z
The summary the script wrote 15 seconds after the alarm went active. A CSV with the 674 samples was saved beside it.

The same approach sends a report, opens a ticket, or posts to a channel. For pages and phone calls, the built-in notification pipelines already deliver by email, webhook, SMS, and voice.

Journal & current state

Alarm transitions in files; current alarms through the API.

The JSON-lines journal records alarm transitions with alarm and project identity, priority, state, occurrence, timestamp, and operator attribution where applicable. Project HTTP and WebSocket APIs expose current alarms and changes.

Current alarms

Read your project’s current alarms over HTTP.

Live changes

Subscribe to alarm events through the project alarm WebSocket endpoint.

Acknowledgment state

View acknowledgment state and the operator attribution recorded for an alarm occurrence.

Notification grouping

Group notification instances by source, occurrence, or configured area.

File journal

Review recorded transitions in logs/alarms.jsonl alongside the site’s operational logs.

Bounded shelving

Shelving has a configured duration and automatically ends when that duration expires.

Alarm journals are part of your operational recovery data. Bring your alarm requirements to a technical review.

Want to talk through your alarm rationalization plan?

Bring your alarm list and a few problem tags. We'll model deadbands, pipelines, and rosters with you on the demo gateway.