Look at sources and accumulation conditions, not the entire steelworks

Steel sites combine high-temperature equipment, fuel gas, reducing gas, welding shielding gas, and several kinds of confined spaces. A blanket rule such as “add CO₂ as the fifth gas in every steel operation” is therefore inaccurate. A CO₂ channel should be added only after determining, in order, whether CO₂ is actually generated or supplied by the process, whether a leak can travel to an occupied area, and whether it can accumulate if ventilation stops.

CO₂ is colorless, odorless, and nonflammable. NIOSH recommends occupational exposure limits of 5,000 ppm as an 8-hour time-weighted average, 30,000 ppm as a short-term limit, and 40,000 ppm as the concentration immediately dangerous to life or health, or IDLH. These are U.S. values and do not replace Korean statutory limits, but they show why CO₂ must not be treated merely as an indoor-air-quality reading or an indirect indicator of oxygen displacement. Because exposure can cause headache, dizziness, breathing difficulty, loss of consciousness, and even convulsions, direct measurement is necessary where generation is possible.

Two broad conditions raise the priority of a CO₂ channel at a steel site. The first is a clearly identified CO₂ source, such as blast-furnace or basic-oxygen-furnace byproduct gas, combustion exhaust, shielding gas, fire-extinguishing agent, or dry ice. The second is human entry or occupancy in a place where gas can remain, such as a pit, duct, tank, furnace interior, basement, or temporary sound or dust enclosure. The more these conditions overlap, the stronger the case for combining fixed monitoring, remote pre-entry testing, and personal monitoring during the job.

1. Blast furnaces, basic oxygen furnaces, and byproduct-gas systems

Integrated steelmaking processes handle gas streams with substantial CO₂ concentrations. World Steel Association material on carbon capture gives approximate ranges of 20~25 mol% CO₂ for blast-furnace gas, 15~20 mol% for basic-oxygen-furnace gas, 15~30 mol% for blast-furnace stoves, and 10~20 mol% for coke-oven flues. These are not workplace-air concentrations. They mean that process gas inside piping, ducts, gas holders, and dust-collection and recovery systems can become a major release source if containment is lost.

It is especially dangerous to look only at CO₂ in these areas. A historical NIOSH blast-furnace investigation reported that blast-furnace gas could contain 27~30% CO, and direct-reading CO instruments exceeded their 100 ppm measurement range at several locations during the survey. Process conditions and equipment differ among facilities, but there is no basis for adding CO₂ while omitting CO from the acute-risk assessment of blast-furnace and basic-oxygen-furnace gas systems. If the gas composition and operating conditions allow flammable constituents, separate LEL monitoring is also needed.

The priority candidates for CO₂ monitoring are therefore worker exposure pathways, not the inside of piping where gas is normally contained. Investigate gas holders and compressor rooms, blower rooms, valve stations, analyzer rooms, drain ports, condensate pits, duct access openings, areas around dust collectors, carbon-capture and compression equipment, and furnaces or vessels opened for maintenance. Review recurring leak records, purge and blowdown paths, dispersion modeling after an exhaust-fan shutdown, and actual airflow together.

2. Combustion areas such as reheating furnaces, heat-treatment furnaces, and boilers

Complete combustion of carbon-containing fuel with sufficient oxygen produces CO₂, while incomplete combustion can increase CO. Representative equipment includes slab reheating furnaces, heat-treatment furnaces, drying ovens, boilers, burners, emergency generators, and internal-combustion equipment. CO₂ normally discharged through a stack does not by itself mean workplace monitoring is needed. There must be a credible scenario in which exhaust enters the work area, such as a flue leak, backdraft, burner fault, damper malfunction, exhaust-fan shutdown, loss of negative pressure, or closed access openings in winter.

Here CO₂ and CO answer different questions. Rising CO₂ can indicate the entry of combustion products and inadequate ventilation, while CO directly indicates the toxic hazard of incomplete combustion. Low CO₂ does not mean CO is absent, and low CO does not mean CO₂ exposure is small. Fuel leaks and explosion potential belong to the LEL channel. Even when CO₂ is added around combustion equipment, it must not replace CO and LEL monitoring, burner interlocks, combustion control, or local and general ventilation.

A single CO₂ reading in the center of a broad, open rolling bay may not be representative. In contrast, a small leak can become an accumulated concentration in an underground utility room, inspection pit, space under a furnace, air entering a crane cab, temporary maintenance tent, or acoustic enclosure. To decide whether permanent fixed monitoring is warranted, measure startup, shutdown, emergency shutdown, and maintenance states separately from normal operation.

3. CO₂-shielded welding and cutting

Steel fabrication and repair may use pure CO₂ or a mixture of argon and CO₂ as shielding gas for gas metal arc welding. OSHA welding guidance lists argon, helium, nitrogen, and CO₂ as shielding gases and explains that CO and CO₂ may also be generated during the process. In confined or partially enclosed spaces, these gases can displace air and create an asphyxiation hazard.

The fact that a small amount of welding occurs in an open fabrication bay does not mean every worker needs a CO₂ sensor. The situation changes, however, inside tanks, ducts, pipes, or furnaces; in vehicle inspection pits, ship blocks, box girders of large structures, weld areas surrounded by wind screens, or poorly ventilated robotic welding cells. CO₂ can accumulate rapidly when a cylinder or manifold hose leaks, a valve remains open after work, several torches operate at once, or local exhaust stops.

In these cases, combine direct CO₂ measurement with O₂ measurement and assess the CO that welding may generate. Metal fumes and other hazardous gases arising from coatings, plating, cleaning agents, and base-metal composition cannot all be addressed by four or five channels on a multigas instrument. Place local exhaust close to the source and manage work permits and substance-specific industrial-hygiene assessments separately.

4. Inerting, purging, and fire-extinguishing systems

Steel sites may use inert gases such as nitrogen or argon to prevent fire or explosion, inhibit oxidation, preserve equipment, or prepare for maintenance. The first step is to confirm what substance is actually introduced. If a system is purged with nitrogen or argon, adding a CO₂ sensor will not directly detect a leak of that gas. O₂ is the key channel in that case, and supply isolation, ventilation, and entry control must work with it.

If CO₂ itself is used for purging, inerting, or as a fixed extinguishing agent, a CO₂ channel has direct value. Review cylinder rooms, manifolds, areas around discharge nozzles, protected-area entrances, adjacent low points, and penetrations between cable basements and control rooms as potential leak paths. Alarms should be integrated with the pre-discharge delay and evacuation signal, restart of ventilation, access locking, and equipment-status indication. A monitor neither prevents discharge nor automatically rescues people, so independent escape means and emergency procedures are required.

5. Jobs that introduce dry-ice cleaning

Dry-ice blasting may be selected to remove contamination from dies, rolls, electrical equipment, and production-equipment surfaces without water or residual abrasive media. Not every steelworks uses this method, so first confirm purchased items and outsourced maintenance methods. OSHA technical guidance describes the advantage that dry-ice pellets sublime after striking the surface, leaving little separate blast-media waste, but warns that they can readily create an oxygen-deficient atmosphere in a confined space.

Dry ice weighing 1 kg does not remain solid; all of it becomes gaseous CO₂. Use the amount consumed, job duration, space volume, mechanical ventilation rate, and discharge location to calculate a worst case, then confirm it through actual measurement. Continuous simultaneous monitoring of O₂ and CO₂ is appropriate inside furnaces, tanks, underground pits, dust enclosures, and enclosed cleaning booths. If coating, scale, or dust from the cleaned surface becomes airborne, it requires a separate exposure assessment.

Do not confuse background concentration with an occupational-safety alarm

Fresh outdoor air contains roughly 0.04% CO₂, and occupied offices develop higher concentrations because of breathing. HSE guidance says CO₂ can serve as a broad indicator of ventilation in ordinary occupied spaces, but explicitly notes that it may be unsuitable or unrepresentative where a process generates CO₂ or in large production buildings where air is not fully mixed. Office ventilation-management values such as 1,000~1,500 ppm must therefore not be copied directly as toxic-exposure alarm values for a steelworks.

Establish a site baseline by recording clean outdoor air and concentrations at multiple locations during normal operation. Also observe changes by shift, season, door position, fan operation, and operating rate. Then set alarms from applicable Korean law, site exposure limits, the emergency-response objective, and the sensor range and error. A level slightly above background may help trace a source, but it is not equivalent to a health-hazard alarm. Conversely, a low average can miss a short, large release such as a blowdown or cylinder leak.

Why O₂ alone cannot stand in for CO₂

CO₂ displaces air, but an O₂ alarm alone may not warn early enough about the harmful effects of CO₂ itself. As a simplified illustration, suppose pure CO₂ mixes into fresh air at 4% while displacing the same volume of air. Oxygen would be about 20.1%. That can remain above the commonly used oxygen-deficiency alarm point of 19.5%, while CO₂ at 4% equals the NIOSH IDLH value of 40,000 ppm. Real mixing is more complex because of temperature, pressure, airflow, and reactions, but the conclusion remains: the two sensors do not measure the same hazard.

CO and LEL cannot replace CO₂ either. A CO sensor addresses the toxic risk from incomplete combustion and CO-containing byproduct gas. An LEL sensor indicates whether a flammable gas is approaching an ignitable range. CO₂ is nonflammable and does not appear as an LEL reading. The four channels can be distinguished as follows:

  • O₂: checks air displacement, oxygen consumption or enrichment, and the reading conditions of some sensors.

  • CO: checks poisoning hazards from incomplete combustion and CO-containing byproduct gas.

  • LEL: checks the fire and explosion potential of fuel gas and flammable constituents.

  • CO₂: directly checks CO₂ leakage, accumulation, and exposure level.

CO₂ is therefore not a replacement sensor that removes existing O₂, CO, or LEL channels. It is a supplemental sensor that fills a gap identified by the risk assessment. If other hazards are expected, such as H₂S, SO₂, NO₂, ammonia, solvent vapor, or metal fumes, they also require appropriate separate measurement.

Sensor placement does not end with “it is heavy, so put it near the floor”

CO₂ has a greater relative gas density than air, so it is reasonable to examine accumulation in pits and low areas. Hot exhaust can rise, however, and high-pressure leak jets, forced airflow, opening doors, thermal convection, and duct suction change the concentration distribution. A row of sensors only at floor level can miss the actual leak path or the worker’s breathing zone.

Choose fixed-sensor locations by combining the expected source, the space between the source and people, occupiable low points, ventilation dead zones, worker routes, and work positions. In a gas-compressor room, assess seals and flanges with entry routes; in a welding cell, the torch and gas manifold with worker positions; in a dry-ice booth, the feed and exhaust points with the recovery waiting area. A point directly in front of a supply-air inlet may understate the hazard through dilution, while a point immediately beside a strong exhaust may be biased above or below the room average, so the monitoring objective must be explicit.

Before entering a confined space, use pumped remote sampling to check the upper, middle, and lower levels and the path of travel separately. OSHA atmospheric-testing procedures for confined spaces likewise say that, where stratification is expected, the direction of travel and surrounding atmosphere should be tested in stages while respecting the manufacturer’s minimum response time at each point. Add sample transport time when using a long hose. If the work can change conditions after entry, place a personal monitor in the worker’s breathing zone or maintain continuous monitoring.

Sensor technology and range matter as much as location. Nondispersive infrared, or NDIR, technology is widely used for CO₂, but that does not make a low-range indoor-air-quality instrument suitable for high-concentration occupational-safety alarms. Confirm the expected worst-case concentration, resolution, upper limit, response time, effects of temperature, humidity and pressure, suitability for hazardous locations, and the calibration and functional-test methods. After a high-concentration exposure drives an instrument beyond its measurement range, follow the manufacturer’s procedure to confirm its condition before reuse.

A site review sequence for deciding whether to add a channel

When adding a CO₂ channel, it is safer to start with process information rather than an equipment catalog. The following sequence can reduce both overinstallation and blind spots:

  1. List substances and gas compositions in blast furnaces, basic oxygen furnaces, reheating furnaces, boilers, welding, purging, extinguishing, and cleaning processes.

  2. Separate release quantities and paths for normal operation, startup and shutdown, blowdown, cleaning, maintenance, and ventilation failure.

  3. Connect pits, ducts, tanks, furnace interiors, basements, temporary enclosures, and worker breathing zones through drawings and field smoke tests or wind-direction checks.

  4. Mark substances not detected by existing equipment among O₂, CO, LEL, and substance-specific toxic-gas channels.

  5. Distinguish the purposes of fixed and portable monitoring, pumped pre-entry testing, personal monitoring, and industrial-hygiene sampling.

  6. Test evacuation, supply isolation, forced ventilation, reentry approval, and rescue procedures after an alarm.

  7. Include bump testing, calibration, sensor life, data review, and corrective action for recurring alarms in operating procedures.

At a steel site, a CO₂ channel ultimately belongs not simply “where CO₂ seems likely,” but where a pathway for CO₂ release and human exposure has been demonstrated. Representative candidates include process-gas systems, enclosed combustion areas, CO₂-shielded welding, CO₂ purging and extinguishing, and dry-ice cleaning. O₂, CO, LEL, and other hazardous agents must be considered alongside it at each location. Adding one sensor can be useful, but genuine safety arises only when source isolation, ventilation, work permits, measurement locations, and alarm response operate as one system.

Sources

Cover photo

Original photo on Pexels