Authorized Gas Tester (AGT) common mistakes
Authorized Gas Tester (AGT) · Understand the traps and how to avoid them.
Treating a 'zero' reading on a combustible gas (LEL) meter as proof that no flammable vapour is present, and issuing the entry or hot work certificate on that basis without cross-checking with a second technology or a colourimetric tube.
Catalytic bead (pellistor) sensors can be poisoned or desensitised by silicones, sulphur compounds, leaded compounds and halon, and they respond poorly or not at all to many high-flashpoint solvents with low vapour pressure. Candidates learn 'LEL meter = flammable check' as a single step and never build the habit of asking what the sensor cannot see.
What to do instead
Before signing, run a two-question check: (1) what is the contaminant likely to be, and (2) does my sensor respond to it? If the substance is a known pellistor poison or a low-vapour-pressure solvent, confirm with a colourimetric detector tube or a PID and record both readings. State the substance and the confirmation method on the certificate.
Bump testing once at the start of the shift and then treating the instrument as 'tested' for the rest of the day, including after a knock, a dropped unit, a long period in a dusty or solvent-laden atmosphere, or a change of gas cylinder.
Candidates conflate the annual calibration certificate with a pre-use functional check, and conflate a bump test with a calibration. They assume one pass covers the shift, so the instrument's response is never re-verified when conditions that could degrade it have occurred.
What to do instead
Learn the three distinct activities and when each applies: calibration (against a traceable standard gas, on a schedule, records the response factors), bump test (brief exposure to gas to confirm sensor and alarm response, before each use or per site procedure), and functional check (battery, pump flow, filter, alarm sounder, display, zero in clean air). Repeat the bump test after any event that could affect the sensor: a drop, a blocked filter, a suspected exposure to poisons, or a cylinder change.
Zeroing the instrument in the confined space or in the work area itself, rather than in clean ambient air, so a contaminated atmosphere is set as the reference point and the subsequent readings are falsely low.
Candidates see the 'fresh air zero' prompt as a formality to clear before they start and press it wherever they happen to be standing, often at the tank hatch where solvent vapour is drifting. They do not connect the zero step to the measurement baseline.
What to do instead
Zero in air known to be clean, outside the space and upwind, and verify with a reading on a second point or a spare instrument if there is any doubt. If the instrument cannot be zeroed cleanly, do not use it - treat that as a failed functional check and take it out of service. Record where and when the zero was performed.
Testing only at the entry point or the hatch and then declaring the whole space safe, without testing at the bottom, at the top, in remote corners, behind baffles, or in adjacent connected compartments.
Candidates picture the space as a single well-mixed volume. In reality gases stratify by density (methane and ammonia rise, hydrogen sulphide and carbon dioxide collect low), and dead legs, sludge, scale and residue release vapour slowly after the space is opened. The candidate also underestimates how much the act of opening and ventilating changes the picture.
What to do instead
Plan sample points from the space geometry and the likely contaminant: top, middle and bottom, plus any remote or poorly ventilated zone and any connected line or compartment. Use an aspirated instrument with sufficient sample line length and allow the full purge time so the reading reflects the point, not the previous one. Note every sample point and depth on the certificate.
Accepting a single set of readings as valid for the whole job, and not re-testing after a break, a shift change, a change in weather, the start of hot work, or any interruption in ventilation.
Candidates treat the entry certificate as a static permission rather than a snapshot. They do not build the mental model that a confined space is a dynamic system: sludge can be disturbed, residues can off-gas as temperature rises, and ventilation can fail silently.
What to do instead
Attach an explicit re-test interval and trigger list to the certificate: at defined intervals, on each re-entry, after any break, after ventilation stops, when the work changes, and continuously where the risk justifies it. Where continuous monitoring is used, make sure the person inside can see and hear the instrument and that someone outside knows the alarm response.
Assuming that because oxygen is at 20.9% the atmosphere is safe, and not recognising that a normal oxygen reading can coexist with a lethal concentration of a toxic gas or an explosive concentration of a flammable gas.
Oxygen is the first number candidates look at, and a normal figure feels like an all-clear. Candidates also miss the reverse error: enriched oxygen (above about 23.5%) dramatically increases the flammability of clothing and materials and is a hazard in its own right, not a safety margin.
What to do instead
Always read and record all four channels (oxygen, flammable, and the relevant toxics) and evaluate each against its own limit - never let one channel vouch for another. Learn the two oxygen failure directions separately: deficiency (displacement or consumption, below about 19.5%) and enrichment (leaks from oxygen lines or cutting sets, above about 23.5%), and the different controls each demands.
Testing for flammable gas with an LEL meter before hot work but not converting the reading correctly for the gas actually present, and not accounting for the fact that many meters are calibrated on methane or a stated reference gas.
Candidates read the percentage LEL straight off the display and compare it to a permit limit without asking which gas the instrument was calibrated on and whether a response factor applies. For heavier hydrocarbons and solvents, the displayed figure can be misleadingly low or high.
What to do instead
Identify the reference gas on the calibration certificate and apply the correct response factor or correction for the target gas, or use an instrument calibrated for that gas. State on the hot work certificate which gas the instrument was set up for, the reading obtained, the correction applied and the permitted limit, and confirm the limit against the site's hot work standard rather than memory.
Confusing the roles on the permit - assuming the gas tester also authorises the work, or that the permit itself makes the space safe, rather than understanding that the tester supplies measurements and the authorising person makes the decision to permit.
The roles overlap in practice on small sites, and candidates answer from their own workplace habit rather than from the defined responsibility split. They also treat the permit as a guarantee rather than a record of controls that must still be verified in the field.
What to do instead
Learn the split explicitly: the gas tester performs and records the tests and states the results and their validity; the authorised/issuing person decides whether to issue the permit and defines its conditions; the supervisor controls the work; the entrant and the standby/attendant perform their own defined duties. Be able to say who does what at each stage - isolation, testing, issuing, entry, monitoring, suspension and cancellation.
Treating the standby person or attendant as a spare pair of hands who can enter the space to help, and planning rescue as 'call the emergency services' with no site-specific arrangement for retrieval.
Candidates underestimate how quickly atmosphere-related casualties occur and how often would-be rescuers become the second casualty. They also assume external responders will arrive in time and know the site, when in practice most confined space fatalities involve untrained rescuers entering without protection.
What to do instead
State the rule clearly: the attendant does not enter, and rescue is by trained personnel using the planned method - retrieval line and harness where the geometry allows, breathing apparatus and trained rescuers where it does not. The plan must cover communications, raising the alarm, isolating the space, access for responders and a practised drill, and must be in place before entry, not written after the incident.
Treating risk assessment and the safe system of work as paperwork to be completed after the decision, rather than deriving the test regime, the permit conditions and the emergency arrangements from identified hazards.
Candidates learn the documents as forms with fields to fill, so the assessment becomes a retrospective justification for a decision already made. The link between 'what could go wrong here' and 'what I therefore test for, how often, and with what' is never built.
What to do instead
Practise working forwards: list the hazards the space and the task create (residue, sludge, connected lines, adjacent work, hot work ignition sources, oxygen enrichment), then derive from each the specific test, the instrument, the frequency, the control and the emergency provision. If a control on the permit cannot be traced back to a hazard in the assessment, either the assessment is incomplete or the control is unnecessary - find out which.