Gas Testing on Deck – What the Numbers Mean

The meter does not make the space safe. It tells you whether the space is trying to kill you right now.

Introduction

Every enclosed space fatality investigation follows the same arc. Someone opened a hatch, someone looked inside, someone went in. The meter was on the bridge, or in a drawer, or had been used hours earlier and showed green. Or it was used at the opening and nowhere else. Or the reading was taken, noted on a permit, and then the atmosphere changed while no one was watching.

The gas meter is the only instrument that stands between a routine task and a body recovery. It is not a formality. It is not a box on a checklist. It is a safety-critical instrument with calibration requirements, sampling limitations, and interpretation demands that most crews never fully grasp – because no one teaches them what the numbers actually mean in context.

This article is not about how to switch the unit on. It is about what the readings represent, why a single clean result means almost nothing, and where the common failures hide.

Contents

• 1. The Instrument Itself – What It Is and What It Is Not
• 2. Calibration: Bump Tests, Full Calibration, and the Records That Matter
• 3. The Four Gases and Their Thresholds
• 4. Sampling: Where, How, and How Many Points
• 5. Interpreting the Numbers – Reading, Trend, Pattern, Source
• 6. One Clean Reading Is Not the End of Testing
• 7. Common Failures That Kill
• 8. Closing Reality

1. The Instrument Itself – What It Is and What It Is Not

A portable multi-gas detector is a precision analytical instrument. It contains electrochemical cells, catalytic bead sensors, or infrared detectors – each with a defined lifespan, each sensitive to cross-contamination, each capable of drifting out of tolerance without any visible warning.

It is not a torch. It is not a gadget. It does not belong in a pocket with spanners and rags.

The cells degrade over time whether the unit is used or not. A catalytic LEL sensor poisoned by silicone vapour will read zero in a space full of hydrocarbon gas. An electrochemical H2S cell past its service life will under-read or fail to respond at all. There is no alarm for sensor death. The screen still lights up. The unit still powers on. It simply lies.

Every meter has a stated sensor life, typically 24 to 36 months for electrochemical cells, less for catalytic bead sensors in harsh environments. That life starts from manufacture, not from the date it was taken out of the box. Ship operators who buy meters in bulk and store them for years before deployment are starting with degraded sensors and no way of knowing it unless they calibrate properly.

2. Calibration: Bump Tests, Full Calibration, and the Records That Matter

There is persistent confusion between a bump test and a full calibration. They are not interchangeable.

A bump test – sometimes called a function check – exposes each sensor to a known concentration of target gas and confirms the sensor responds. It verifies that the alarms trigger. It does not adjust the reading. If the sensor reads 15 ppm when exposed to 25 ppm of test gas, the bump test tells you the sensor is alive but does not correct the error. That is the limit of its value.

A full calibration exposes the sensor to known gas concentrations and adjusts the instrument’s output to match. It is a zeroing and spanning exercise. It requires certified calibration gas within its expiry date, a calibration adapter, and a competent operator who records the before-and-after readings.

Bump tests should be performed before every use. Every single time. Not weekly. Not when someone remembers. Before every entry, every permit, every atmospheric assessment.

Full calibration follows manufacturer intervals – typically every 180 days – but must also be performed whenever a bump test shows deviation beyond the manufacturer’s stated tolerance, after sensor replacement, or after any exposure to high concentrations of target gas.

Records matter. Every bump test result and every calibration record must be logged with the date, the gas concentrations used, the pre- and post-calibration readings, and the identity of the person who performed it. If the records do not exist, the calibration did not happen. Auditors, investigators, and coroners all take that view.

A meter without a calibration record is an ornament.

3. The Four Gases and Their Thresholds

Standard four-gas meters measure oxygen, flammable gas as a percentage of the Lower Explosive Limit, hydrogen sulphide, and carbon monoxide. Each has threshold values that determine whether entry is permitted, and those thresholds are not optional guidance – they are hard limits.

Oxygen (O2)

Normal atmospheric oxygen is 20.9%. The safe working range is 19.5% to 23.5%.

Below 19.5%, the space is oxygen-deficient. Judgment and coordination begin to fail around 16%. Below 12%, unconsciousness occurs within seconds and without warning. There is no sensation of suffocation. The body does not recognise oxygen depletion the way it recognises CO2 build-up. A person in a 6% oxygen atmosphere takes one breath and drops.

Above 23.5%, the space is oxygen-enriched. Clothing, hair, and materials that would not normally ignite become flammable. Sparks that would be harmless in normal air become ignition sources. Oxygen enrichment is less commonly encountered on deck but occurs around leaking oxy-acetylene equipment, poorly ventilated paint stores, and certain cargo operations.

Lower Explosive Limit (LEL)

The LEL reading shows the percentage of the Lower Explosive Limit present, not the percentage of gas in the atmosphere. A reading of 10% LEL for methane means the atmosphere contains roughly 0.5% methane by volume – one tenth of the concentration required for ignition.

The general threshold for safe entry is below 10% LEL. For hot work, it is typically below 1% LEL. For certain cargoes – particularly those governed by ISGOTT – the threshold may be ‘not more than 1% LEL’ for any entry at all.

The difference between 10% LEL and 100% LEL is the difference between a safe space and a bomb.

LEL sensors are typically catalytic bead type. They require oxygen to function. In an oxygen-depleted atmosphere, the LEL reading may be artificially low or zero even when flammable gas is present. This is a critical limitation that many operators do not understand. If the O2 reading is low, the LEL reading cannot be trusted.

Hydrogen Sulphide (H2S)

H2S is heavier than air, accumulates in lower sections of tanks and spaces, and is immediately dangerous to life at concentrations that are still easily smelled – until it paralyses the olfactory nerve and the smell vanishes entirely. That disappearance is not a sign the gas has cleared. It is a sign the exposure has become more dangerous.

The occupational exposure limit varies by jurisdiction. The common short-term exposure limit is 5 to 10 ppm. At 100 ppm, death can occur within minutes. The meter alarm is typically set at 10 ppm low and 15 ppm high.

On tankers, especially those carrying sour crude or having residues from previous cargoes, H2S concentrations in ballast tanks, pump rooms, and void spaces can reach lethal levels with no external indication.

Carbon Monoxide (CO)

CO is a product of incomplete combustion. On deck, sources include engine exhaust drawn into enclosed spaces, smouldering cargo, and the decomposition of certain chemical cargoes. The TWA exposure limit is 25 ppm in most jurisdictions. The STEL is 50 ppm, though some flag state standards set it lower.

CO binds to haemoglobin with roughly 250 times the affinity of oxygen. Chronic low-level exposure produces symptoms that mimic fatigue and seasickness – headache, nausea, confusion – and are routinely dismissed aboard ship until someone collapses.

4. Sampling: Where, How, and How Many Points

A reading taken at the manhole is a reading of the atmosphere at the manhole. It is not a reading of the atmosphere at the bottom of the tank, or in the far corner, or at the level where someone will be working.

Sampling must be conducted at multiple heights and multiple locations within the space. The minimum is top, middle, and bottom, but the geometry of the space dictates more. Cofferdams, double bottoms, chain lockers, cargo holds with complex framing – each has pockets where gas accumulates and ventilation does not reach.

Hydrocarbons and H2S are heavier than air. They settle. A clean reading at the top of a tank while a lethal concentration sits at the bottom is not a rare scenario. It is the standard condition in an unventilated cargo tank after discharge.

Other gases behave differently. CO diffuses relatively evenly. Oxygen depletion from rusting or biological activity can be localised to areas with the greatest surface contact – the bilge, the lower frames, the spaces behind corroded coatings.

Remote sampling uses a motorised pump drawing atmosphere through a length of tubing. The tubing length matters. A 10-metre sampling line introduces a time delay – the pump must draw a sufficient volume to purge the line before the reading stabilises. Manufacturer data specifies the pump’s flow rate. A typical rate of 0.5 litres per minute through a 10-metre line of 3mm internal diameter means roughly 45 seconds of purge time before the reading at the sensor reflects the atmosphere at the probe tip. Impatient operators who glance at the reading before the line has purged get the atmosphere from the previous sample point, or from ambient air.

If the sampling line was not purged, the reading belongs to a different atmosphere.

5. Interpreting the Numbers – Reading, Trend, Pattern, Source

A number on a screen is the lowest level of information. What matters is what the number means in context.

The hierarchy of interpretation runs: reading, trend, pattern, source.

A single O2 reading of 20.8% is normal. But if the previous reading was 20.9%, and the one before that was 20.9%, and the space has been sealed and ventilated, then a drop of 0.1% over 30 minutes is a trend. That trend indicates something in the space is consuming oxygen – rusting steel, biological activity, a chemical reaction. The rate of consumption may be slow now but will not necessarily remain slow.

A pattern emerges when readings from different sample points are compared. If the lower sampling point consistently reads lower O2 than the upper point, the consumption source is at the bottom and the ventilation is not reaching it. If LEL readings are higher at one end of a tank, the source of hydrocarbon vapour is localised there – residual cargo, sediment, a coating breakdown.

Identifying the source is the final and most important step. A reading tells you what is present. A trend tells you it is changing. A pattern tells you where. The source tells you why – and whether the condition will persist, worsen, or return after ventilation stops.

Officers who record a single set of readings on a permit and treat the matter as closed have completed a bureaucratic exercise, not an atmospheric assessment.

6. One Clean Reading Is Not the End of Testing

Atmospheres change. They change when ventilation is stopped, when hatches are repositioned, when the sun heats a tank and volatilises residual cargo, when work disturbs sediment, when welding consumes oxygen locally, when a pump room bilge is agitated.

A reading taken at 0800 before entry does not describe the atmosphere at 1030 when the work is underway. It describes a historical condition that may no longer exist.

Continuous monitoring – either by personal gas monitors worn by entrants or by fixed instruments at the work site – is the only reliable method. Where continuous monitoring is not available, repeated spot checks at defined intervals are mandatory. The interval depends on the space, the work, and the conditions, but it is never ‘once at the start and done’.

The permit to work should specify the re-testing interval. If it does not, the permit is incomplete.

An atmosphere that tested safe an hour ago is not an atmosphere that is safe now. It is an atmosphere that was safe an hour ago.

7. Common Failures That Kill

The failure modes in gas testing are well documented. They recur because they are rooted in habit, convenience, and misunderstanding – not malice.

Bump testing by exhaling into the sensor

This is more common than anyone in a safety department wants to admit. A crew member breathes onto the sensor. The CO2 in exhaled breath triggers a minor response on some cells, the O2 reading dips slightly, and the operator concludes the meter is ‘working’. It is not a bump test. It proves nothing about the LEL, H2S, or CO sensors. It does not expose any sensor to a known concentration. It is a performance of competence, not a demonstration of it.

Sampling only at the entry point

The manhole, the access hatch, the top of the ladder. This is where the ambient air mixes in. This is the cleanest reading in the space. Lethal concentrations a metre below the opening will not register on a meter held at the coaming. Remote sampling lines exist precisely to eliminate this failure. If the line is not deployed, the assessment is incomplete.

Trusting a pre-work reading as a continuous condition

A reading taken during the morning meeting, written on the permit, and not revisited. The Responsible Officer signs. The entrant descends. Two hours later the atmosphere is different and no one knows because no one checked. This is the single most common scenario in enclosed space fatality reports. The permit becomes a death certificate with a timestamp that proves the last known safe reading was hours before the casualty.

Using a meter with expired sensors or overdue calibration

The meter powers on. The display looks normal. No error is shown. The operator has no reason to doubt it – unless they checked the calibration log and the sensor installation dates. Degraded sensors under-read. They do not fail high. They do not fail to alarm on the safe side. They drift towards zero, towards ‘clean’, towards the reading that lets everyone proceed. This is the most dangerous kind of failure: the instrument that confirms what you want to hear.

Failing to account for the LEL sensor’s oxygen dependency

In an inerted or oxygen-depleted space, the catalytic bead sensor cannot oxidise the target gas. The LEL reads zero. The operator sees a low O2 reading and a zero LEL reading and does not connect the two. The space may be full of hydrocarbon vapour. The moment ventilation introduces oxygen, the atmosphere crosses through the flammable range. If there is a spark at that moment, the ignition occurs in a space everyone believed was gas-free.

8. Closing Reality

The gas meter is the last line of defence before a human being enters an atmosphere that cannot be assessed by any other sense. It cannot be seen, smelled, or felt reliably. The instrument must be maintained, calibrated, deployed correctly, and – most critically – its readings must be understood, not just observed.

A number on a screen is not safety. It is data. Safety is the competence to know what that data means, to know when it is insufficient, and to know when the space needs to be re-tested, re-ventilated, or simply not entered.

A green reading on a neglected meter in the hands of someone who does not understand the limitations is not a safety barrier. It is a false floor over an open hole.