Confined Space Gas Monitoring: O₂, LEL, CO and H₂S Limits and Sensor Selection

Why a confined space is the hardest case for a gas instrument

The definition of a confined space is much the same across jurisdictions: large enough for a person to enter and work in, with limited or restricted means of entry or exit, and not designed for continuous occupancy — tanks, pipes, pits, wells, silos, sewer manholes, vessel holds, reactors. What they share is poor natural ventilation: any leak, residue or reaction product accumulates instead of dispersing the way it would in an open workshop.

Three hazards exist at once. Oxygen deficiency is the most common: microbial action, inert-gas purging, corrosion of steelwork and welding all push oxygen below 19.5%, and between 19.5% and 16% a person often feels only 'a bit stuffy' while judgement and strength are already failing. Among toxic gases, hydrogen sulphide matters most: at low concentrations it smells of rotten eggs, and at high concentrations it paralyses the olfactory nerve — removing the only sensory warning there is. A flammable atmosphere at 10% LEL means a single ignition source can light it, and confined spaces are exactly where hot work, grinding and non-Ex tools are used.

There is a structural factor behind why these incidents so often kill in pairs: restricted access. When the person inside goes down, the attendant's instinct is to climb in and pull them out — and the atmosphere does not become safe just because one more person has entered. That is why regulations make 'the attendant does not enter to rescue' a hard duty: the attendant calls professional rescue, guards the entry point and controls who goes in, rather than going in themselves.

The pass/fail limits you must know by heart

Confined space gas limits are remarkably consistent worldwide; differences live in the detail. The table below sets the four measurements against three main authorities: US OSHA, Singapore WSH (essentially the same numbers as the UK and Australian regimes), and China's GB 30871-2022.

Test sequence and sampling points: test before any part of you enters

Sequence is not a matter of habit. An oxygen level that is too low or too high directly distorts the flammable reading — a catalytic sensor needs oxygen to burn the gas, so it under-reads or goes blind when oxygen is short, and over-reads when oxygen is enriched; some electrochemical toxic sensors also drift at extreme oxygen levels. So the order is fixed: oxygen first, then flammables, then the specific toxics. Singapore's SS 568 and Western practice both follow it.

Sampling points follow gas density. Hydrogen sulphide (MW 34) is heavier than air and pools at the pit floor, tank bottom and lower manhole; methane (MW 16) is lighter and gathers at the tank roof and just under the cover; carbon monoxide (MW 28) is close to air and mixes fairly evenly. Large spaces should be tested top, middle and bottom — and left, centre and right where relevant — not just a probe poke at the opening.

• Use a probe or pumped sampling: draw the sample from depth into the instrument instead of putting your head into the opening. Move the inlet between layers and let each reading stabilise.
• Sampling must be completed within 30 minutes before entry, and entry only after it passes; if work is interrupted for more than 60 minutes, re-test before resuming (GB 30871-2022 clause 6.3).
• During work, keep a portable monitor running continuously at the entry and log readings every two hours (GB 30871-2022 clause 6.5). On alarm, stop work, evacuate and correct — resume only after a fresh test passes.

Choosing the four channels

Confined space instruments are usually built with four channels — O₂, flammable gas, CO and H₂S — because that combination covers the overwhelming majority of fatal atmospheres and matches the minimum test set in most regulations. For a maker of portable instruments or fixed transmitters, the question is not whether the four channels exist but whether each channel's range, resolution and principle actually line up with the pass/fail limits.

The LEL channel: why NDIR suits confined spaces better

The catalytic bead is the most common flammable channel in confined spaces because it is cheap and linear across most hydrocarbons. But two of its preconditions collide head-on with confined space work. First, it relies on the target gas burning flamelessly on the bead, which needs enough oxygen — and confined spaces are precisely where nitrogen purging, inerting or plain deficiency are routine. In such atmospheres the bead under-reads and, in severe deficiency, stops responding altogether.

Second, the bead is permanently poisoned by silicon, lead, sulphur and halogens — and sulphur-bearing gases (H₂S, mercaptans), silicone sealants and paint solvents are exactly what confined spaces contain. Once poisoned, the reading may simply drift low with no outward sign, until the day it is needed to alarm. An NDIR channel does not burn the gas, does not need oxygen and cannot be poisoned, and it can run at milliamp-level power for portables and battery nodes. For purged or inerted tanks and pipelines, NDIR is close to the only dependable choice.

Portable, pumped, fixed: each form factor covers a different phase

Confined space monitoring happens in phases and each phase needs a different form factor — mixing them up is a common mistake. Before entry you need a pumped instrument or a probe: sample from depth while you stay outside. During entry you need a monitor worn on the body, kept on for the whole period, never pocketed, never switched off on the strength of an earlier pass. For spaces that need continuous cover (sewer pump rooms, tank farms, silos) fit fixed detectors interlocked with ventilation.

• Wear it at breathing-zone height — collar or shoulder — not in a pocket or tool bag, where it measures the air next to your trousers.
• Runtime decides form factor: low-power sensing (milliwatt-class NDIR, for example) keeps portables and wireless nodes running a full shift, removing the monitoring gaps that battery swaps create.
• Log everyone in and out: the attendant records entries and exits at the opening, reconciles the headcount and confirms the space is empty at the end.

Calibration and bump tests: half of your compliance record

An instrument that has not been functionally checked is just producing numbers. A bump test briefly exposes the sensors to a known concentration to confirm they respond and the alarms fire; calibration adjusts the reading to the true value and leaves a traceable record. The first proves it will alarm, the second proves it reads correctly, and neither replaces the other. South Africa's mining guideline for confined spaces defines the bump test explicitly and expects it in the record; most manufacturers call for a bump test before every use and periodic calibration matched to how hard the instrument works.

Sensor life and cross-sensitivity belong in the purchase specification too. Electrochemical toxic sensors typically last two to three years and cross-respond (a CO channel reacts to hydrogen, for example); electrochemical oxygen sensors deplete over time and need periodic verification against 20.9% air; infrared channels last long but suffer from contaminated optical windows. Instruments that live with high humidity, dust and sulphur gases usually fall short of datasheet life, so spares and calibration gas belong on the same bill of materials.

Southeast Asia and South Africa: compliance differences and what that means for exported instruments

Singapore is the most codified market in Southeast Asia. Its Workplace Safety and Health (Confined Spaces) Regulations 2009 require an authorised manager to issue the entry permit, and six conditions must be satisfied first — three of them being oxygen within 19.5%–23.5%, flammable gas below 10% of the LEL and toxic substances at or under the legal exposure limits. The permit needs three separate signatures (supervisor, safety assessor, authorised manager), which are not to be combined in one person, and testing follows SS 568. Breaches attract stop-work orders and public enforcement records, which weigh directly on a contractor's project eligibility.

South Africa takes a different route. Under section 9(2) of the Mine Health and Safety Act, the Chief Inspector of Mines can require a mine to prepare and implement a mandatory Code of Practice, and the Mine Health and Safety Inspectorate's guideline on the management of working in confined spaces at mines (DMR 16/3/2/4-B4) lists what that Code must cover: a register of confined spaces, permit systems, energy isolation, atmosphere testing and monitoring, ventilation, a standby person, a rescue plan, record keeping and system review. The guideline defines the bump test as sensor verification and expects it in the procedure. For suppliers to mines and their contractors that means the instrument is only part of the offer — calibration service, record templates and training material travel with it.

• Design to the strictest tier: China caps oxygen at 21% (with 23.5% as the enrichment ceiling), while the international figure runs to 23.5%. If one instrument must serve both, make the display and alarm logic configurable per region rather than hard-coding a 21% ceiling.
• Equipment used in hazardous areas must match the required explosion protection and local certification; confined spaces are frequently explosive atmospheres, so enclosure, certification and the intrinsically safe circuit belong in the same selection decision as the sensor.
• Keep spares and calibration gas locally available: cells, sensors and gas are recurring consumables, so build annual replacement volume and calibration service into the total cost — closer to reality than unit price alone.

Frequently asked questions