Process Safety · Lesson 44 · 8 min read

A Fixed Gas Detector Only Protects When the Whole Function Works

A fixed gas detector is not a complete safeguard by itself. Protection depends on the release reaching a suitable sensor, the sensor responding, the signal reaching the logic system, alarms and automatic actions occurring as designed, and people knowing what to do. A green icon or a recent sensor calibration cannot prove that this whole chain will work on demand.

The detector head is only the first link

A flammable-gas release can form a hazardous cloud before anyone can see or smell it. Fixed detection is intended to recognise a developing event early enough for defined responses. But detection performance depends on the release location, gas behaviour, ventilation, congestion, detector technology, field of view or sampling path, alarm settings, voting logic and the actions connected to the signal.

A detector can be healthy yet poorly placed. A release can pass around it because buoyancy, momentum, wind and equipment geometry change the cloud path. A well-located detector can still fail if its sensor is poisoned, obstructed or out of calibration; power is lost; logic has changed; or the linked alarm, ventilation or shutdown action is unavailable.

Official incident: the installed system was not functional

The U.S. Chemical Safety Board investigated the 24 January 2020 explosion at Watson Grinding in Houston. A degraded, poorly crimped hose disconnected and released propylene overnight. The storage-tank manual valve had not been closed, so gas continued to enter the coating building. An employee likely ignited the accumulated vapour when entering and turning on lights. Two employees and a nearby resident died, two other employees were injured, and hundreds of structures were damaged. [1]

The coating booths had gas detectors, but the automated gas-detection alarm, exhaust-fan startup and gas-shutoff system was not functional. CSB found that the system had been disconnected from the programmable logic controller and that written concerns raised in 2013, 2016 and 2019—and again two weeks before the explosion—were not effectively addressed. There was no testing programme to assure the functional integrity of the complete system. CSB concluded that, had it worked, the system should have alarmed, started exhaust ventilation and remotely isolated the propylene supply, likely preventing the dangerous accumulation. [1]

This onshore incident is not an offshore detector-layout case. Its transferable lesson is that installed hardware and local calibration do not prove the complete protection function.

Hypothetical FPSO example: a partial test gives partial assurance

Hypothetical example—not a documented incident: During a planned FPSO test, a technician applies the approved test gas to a detector near a compressor module. The detector head responds and the control-room display changes, but the authorised end-to-end test shows that a configured automatic action does not occur. A previous modification had left the signal mapped differently from the current cause-and-effect document.

The team stops the test under the approved plan, declares the affected function impaired, applies authorised operating restrictions and investigates through the installation's change and defect processes. They do not treat the responsive sensor as proof that the barrier is available. This example supplies no universal alarm set point, detector spacing, voting arrangement or permission to test live systems.

Define what the safety function must achieve

HSE's offshore fire-and-explosion guidance emphasises timely detection and the capabilities of detection arrangements. It also links major-accident measures to performance standards. For a fixed gas-detection function, the installation should be able to state what hazardous scenarios it addresses, where and how quickly detection is required, which alarms or automatic actions follow, what availability and survivability are required, and how achievement is verified. [2]

Keep that basis connected to the current plant. Changed ventilation, obstructions, fluids, operating modes or logic can alter gas dispersion or the response chain. Review them through the applicable change and fire-and-gas arrangements.

Critical controls

Coverage based on credible releases. Use competent dispersion and consequence assessment, detector technology limitations, ventilation conditions and field verification. Do not substitute detector count for demonstrated coverage.

End-to-end proof. Test enough of the chain to verify the claimed function: sensor response, indication, alarm, logic, voting, automatic action and operator response, as applicable. A bump or calibration test proves only what it actually exercises. Plan tests so they do not create an unintended trip or mask genuine hazards.

Impairment control. Identify every inhibited, bypassed, failed, obstructed or overdue device. Assess the affected major-accident scenario, authorise compensating measures and operating limits, communicate them at handover, and track restoration to a verified close.

Actionable alarms and response. Operators need a clear alarm meaning, reliable location information and an approved response. The alarm should prompt safe protective action—not an unplanned search for a leak inside a potentially hazardous area.

Leadership actions

Leaders should ask for barrier evidence, not only the percentage of detector heads marked healthy. Review overdue end-to-end tests, long-standing inhibits, repeated faults, coverage assumptions and discrepancies between cause-and-effect documents and the running configuration. Provide time and specialist support to restore degraded functions, and make production decisions against the barriers actually available.

Three practical actions

• Trace one function: choose a credible release and follow its detector, signal, logic, alarm, automatic action and operator response from field to final element.
• Check impairments: confirm every override, fault and overdue test has an owner, assessed consequence, authorised controls and restoration deadline.
• Challenge change: verify that recent equipment, ventilation, layout, software or process changes were assessed for fire-and-gas coverage and cause-and-effect impact.

Official video: No Detection

The official CSB video reconstructs the release and explains how process-safety and emergency-preparedness weaknesses allowed it to become fatal. View the official CSB video page.

Takeaway and team discussion

Key takeaway: a detector on the wall is evidence of installation; only verified coverage and end-to-end performance provide evidence of protection.

Discussion question: which gas-detection function on our installation has the weakest current evidence that the sensor, logic, final action and human response will all work together?

Continue in the Process Safety hub, and connect this lesson with alarm response, bypass control, management of change and shutdown-system resilience.

Verified primary references

1. U.S. CSB: Fatal Propylene Release and Explosion at Watson Grinding, final investigation report—official findings on the release, disabled detection/alarm/ventilation/shutoff system, mechanical integrity and emergency preparedness.
2. UK HSE: Prevention of fire and explosion, and emergency response on offshore installations (L65)—official offshore guidance addressing timely detection, detection-system capabilities and performance standards.
3. U.S. CSB investigation page: Watson Grinding Fatal Explosion and Fire—official incident summary, report and recommendations.
4. U.S. CSB official video: No Detection—Explosion at Watson Grinding—official incident animation and safety lessons.

Published and sources checked 26 October 2026. Educational discussion only: apply the installation's approved fire-and-gas philosophy, performance standards, management-of-change, impairment, proof-test, hazardous-area and operating procedures. No universal detector spacing, alarm set point, test interval or permission to inhibit protection is supplied.

Discuss this article in the comments →

Register interest in future HSSE email updates