Skip to content
All use cases
  • manufacturing
  • Safety & Security
  • representative

Flame, Smoke & Chemical Leak Micro-Detection in Oil Refineries

A petro-chemical refinery operating hazardous process units, tank farms, and pipe racks monitored by 400 outdoor industrial CCTV cameras.

Runs onArjuna
early flame and smoke detection response time cut
3 minutesearly flame and smoke detection response time cut

How the work runs

The pressure that made this worth automating, the steps the system runs, and what came out the other side.

Pressure & Trigger Points

  • Traditional optical flame and point gas detectors were affected by outdoor wind currents, missing early micro- flames or vapor leaks until fires spread.
  • Early-stage chemical vapor leaks (e.g., steam, hydrocarbon vapors) are invisible to the naked eye but cause major safety disasters if un-detected.
  • Refinery safety rules required 24/7 continuous visual monitoring of high-pressure pump seals.

The run · 5 operational steps

Click any step to inspect telemetry signals, model reasoning, and governance gates.

scroll →

1

AI Video Flame & Smoke Spectral Analytics

Arjuna analyzes CCTV video pixels for characteristic high- frequency flame flickering and subtle smoke plume expansion patterns 10x faster than thermal point sensors.

Input Signal:

Real-time operational telemetry & queue

Reasoning Pattern:

MCP grounded vector inference

Governance Gate:

Policy constrained with audit write-back

Verified Business Outcomes

  • Early flame and smoke detection response time cut from 3 minutes to 4 seconds.
  • Intercepted 3 major high-pressure pump seal vapor leaks before fire ignition occurred.
  • Enhanced refinery fire safety coverage across 400 existing cameras with zero physical sensor piping work.

Capabilities this relies on

  • vision perception
  • anomaly detection
  • alert routing
  • policy compliance
  • in boundary deployment

More in Arjuna