Carbon Monoxide Monitoring at Work: Limits, Alarm Set-Points and Sensor Selection

Why CO is more dangerous than most toxic gases

CO's toxicity is chemical asphyxiation: it binds haemoglobin roughly 200–250 times more readily than oxygen, forming carboxyhaemoglobin (COHb) and robbing red blood cells of their oxygen-carrying capacity. Severity is driven by COHb saturation, not by any sensation — and the body has no sensation for rising COHb.

Symptoms climb with saturation: at roughly 10% COHb, headache and impaired judgement — easily mistaken for fatigue or flu; at 30–40%, severe headache, confusion and collapse; above 60%, frequently fatal within minutes. Because the early symptoms are so non-specific, several workers feeling "flu-ish" at once is often the first sign of a CO event.

The typical accident pattern is remarkably consistent: combustion equipment + poor ventilation + no monitoring. In an October 2020 incident at a coal-gas producer, a maintenance worker entered a coal bunker without gas testing, collapsed, and two colleagues followed him in to help; all three died. Post-incident testing confirmed CO far above the national occupational limit. "No testing, no protection, blind rescue" — those three phrases appear in almost every CO incident report.

Four limit systems and what each one is for

The most common confusion in safety programmes is mixing statutory limits with recommendations. The table below puts the four mainstream systems side by side: China's GBZ 2.1 is the domestic legal basis; the OSHA PEL is the enforceable US limit; the NIOSH REL and ACGIH TLV are health-based recommendations — not legally binding, but routinely cited in industrial-hygiene programmes and customer audits.

• Altitude correction (GBZ 2.1): the MAC drops to 20 mg/m³ at 2,000–3,000 m and to 15 mg/m³ above 3,000 m — don't copy lowland limits onto plateau projects.
• Public-facing indoor standards are far stricter: GB/T 18883 caps indoor CO at 10 mg/m³ (1-hour mean) and the WHO 2021 air-quality guideline at roughly 9 ppm. Occupational limits assume healthy adults on bounded shifts; public standards must protect the elderly, pregnant women and cardiovascular patients.
• Compliance is judged against the legal limit, but designing to the stricter recommendations — ACGIH 25 ppm or NIOSH 35 ppm — builds in margin and protects the more sensitive workers.

How to set the two-stage alarm points

The common design basis in China is GB 50493 for combustible and toxic gas detection: Stage 1 alarm at or below 100% of the OEL, Stage 2 at or below 200%. For CO (anchored to the GBZ 2.1 STEL of 30 mg/m³) that works out to roughly 30 mg/m³ and 60 mg/m³. Stage 1 triggers ventilation and inspection; Stage 2 triggers evacuation readiness.

US practice is essentially the same: a low alarm at 25–35 ppm (conveniently spanning the ACGIH TLV and NIOSH REL) and a high alarm at 50–100 ppm, with high-risk sites pushing the high alarm to 200 ppm, the NIOSH ceiling. EN 50545-1 likewise leaves alarm grading to be adapted to local regulations.

Note the evidential logic of the TWA: an 8-hour time-weighted average is a dose concept, and a single spot reading cannot demonstrate compliance. Where TWA compliance must be proven — occupational hygiene surveys, customer audits — use an instrument with data logging, or sample per the GBZ/T 160.28 method for laboratory analysis.

Sensor selection: the range ladder and the H2 trap

First principle of range selection: the normal alarm point should sit within 10–60% of full scale, and the maximum expected concentration must not overrun the range. The life of an electrochemical CO sensor depends on cumulative exposure — one that routinely works beyond its range fails calibration far faster than one that only sees occasional hits.

• High-concentration zones (blast-furnace and converter gas can run 20–30% CO by volume) demand the high-range tier: a 0–1,000 ppm sensor in a gas area works beyond range daily and drifts beyond credibility within weeks.
• Hydrogen cross-interference is the most overlooked trap: a standard electrochemical CO sensor responds to H₂ at roughly 22% — 100 ppm of hydrogen adds a phantom ~22 ppm to the CO reading, enough to trip a Stage 1 alarm. Wherever CO and H₂ coexist — hydrocracking, methanol and ammonia plants, coke ovens, electrolysis — either specify a low-H2 CO sensor (cross-response below 3%, or <10 ppm at 100 ppm H₂) or add an H₂ sensor to tell real readings from phantom ones.
• Calibration: portable CO instruments are typically calibrated every 6–12 months; fixed systems follow the manufacturer's schedule, shortened for heavy exposure or dusty sites. A bump test before every use is the floor, not the ceiling.

Placement: put the sensor where people breathe

CO's density is close to that of air (slightly lighter), so it mixes with airflow rather than pooling high or low. Fixed detectors belong at breathing-zone height (roughly 1.5–2 m above floor), on the airflow path between the source and where people stay — not hugging the source, where transient puffs cause nuisance alarms. In LPG forklift warehouses the priority is the aisle between loading and charging zones; in car parks, zone by bay density and level; for confined spaces, test before entry and monitor continuously during work.

One last, frequently missed point: wire CO monitoring into the ventilation interlock. Mechanical ventilation sized for peak occupancy but running flat-out year-round is the most common energy waste in warehouses and car parks; demand-controlled ventilation triggered by CO (or CO₂) delivers safety and savings at the same time.

Frequently asked questions