Confined space entry permit
the atmosphere decides, not the size of the opening

A vessel, a sump, a sewer, a silo, a tank car, the space under a floor slab. What makes it dangerous is not the size of the opening — it is that the air inside is not the air outside, and it can change while a person is in there.

What makes a space permit-required?

A confined space entry permit sits inside the wider permit to work system, and it answers one question specific to this hazard: has the space stopped behaving like the air outside it. Four tests, applied in order, decide that. The first three define a confined space; the fourth turns it into one that needs a permit.

That fourth test is where judgement lives. A clean, ventilated pit with no stored energy may be entered under a lighter regime in some jurisdictions. A grain silo, where the material itself can swallow a person in seconds, is permit-required no matter how good the air is.

What are the acceptable atmosphere limits?

Three numbers decide whether a confined space is safe to enter, and all three have to hold at the same time — passing two out of three is not passing. Under OSHA 1910.146 these are the general-industry baseline, and most operators set them stricter, never looser.

Why is oxygen tested before flammable gas?

Oxygen is measured first because the instrument that reads flammable gas cannot be trusted until oxygen is confirmed present. Most combustible-gas sensors burn a small sample of the air against a heated element and report the heat produced as a percentage of the lower explosive limit; that reaction needs oxygen to run correctly, so in an oxygen-deficient atmosphere the same sensor understates the reading rather than failing to respond at all. A space that reads low on both oxygen and flammables has not necessarily passed two tests — it may have failed the first badly enough to hide the second.

That is why the sequence is fixed, not a matter of preference:

1. Calibrate and bump-test the meter against a known gas before the shift starts — an instrument nobody has checked is a guess with a digital display.
2. Take the first reading from outside the space, through the opening or with a probe — putting a head in to get a reading is already an entry.
3. Read oxygen first and confirm it sits within 19.5–23.5% before trusting any other figure the same meter shows.
4. Read flammable gas second, now that oxygen is confirmed adequate for the sensor to respond correctly.
5. Read toxic contaminants third, against the permissible exposure limit for each substance expected in that space.
6. Repeat all three readings at the top, middle and bottom of the space — heavy vapours settle low, light ones collect at the crown, and no single level stands in for the others.
7. Re-test after any interruption — a break in the work, a change in an adjoining system, a shift change — because the permit only vouches for the atmosphere at the moment it was actually tested.

OSHA 1910.146 and equivalent instruments elsewhere expect this order as a matter of course, because the physics behind it does not change with jurisdiction.

Who does what during a confined space entry?

Four roles carry distinct responsibility, and the system depends on none of them being combined. Mixing two of these into one person is one of the more common ways a confined space programme quietly weakens itself.

• Authorised entrant — goes in, wears what the permit specifies, keeps in contact, comes out on the alarm without arguing.
• Attendant — stays outside for the whole entry, keeps count of who is inside, watches the gas monitor and the plant around the space, orders evacuation, calls rescue. Never enters.
• Entry supervisor — verifies the tests, checks isolations, signs the permit, and cancels it when conditions change or the work ends.
• Rescue team — trained, drilled and available in the time the plan assumes, not assembled once something has already gone wrong.

Why must the rescue plan exist before anyone goes in?

The rescue plan has to be complete, staffed and provably workable before the permit is signed, not sketched out once the radio goes quiet. Whatever disabled the entrant — an atmosphere, engulfing material, a fall inside the space — is still present when the rescue starts, so a person who goes in without the same training, the same atmosphere test and the same equipment as the original entrant is not rescuing anyone; they are becoming the second casualty in the same incident.

A workable plan settles each of these before entry, not during it:

1. Retrieval first. Wherever the space and the task allow it, fit the entrant with a harness and a retrieval line run to a tripod or winch at the access point, so a person outside can pull them out without anyone else going in.
2. Where retrieval cannot reach. If the shape of the space — bends, obstructions, a long horizontal run — makes a line useless, name the entry rescue team, confirm its members are trained and equipped exactly as an entrant would be, and confirm they are already on site rather than summoned after the fact.
3. Response time checked, not assumed. Work out, and write down, how long the named rescuers take to arrive, dress and reach the casualty, then weigh that time against the hazard identified for the space. A team stationed some minutes away does not cover an atmosphere that can disable someone in under one.
4. Communication tested in advance. Confirm how the attendant raises the alarm and how the rescue team is alerted, and test that link before the first entry of the job — not during the emergency it exists to answer.
5. Entry rescuers follow the same permit. If the plan calls for rescuers to enter, they test the atmosphere and take the same precautions as the crew they are relieving; the hazard that put the first person down has not gone away because someone else is now in a hurry.

Across decades of confined space fatality data, NIOSH — the US research agency for occupational safety — keeps finding the same pattern: a disproportionate share of the dead are the rescuers, not the original casualty. The permit’s insistence on a proven plan, named people and a tested response time is the control that exists specifically to break that pattern.

How is the space made safe before the first reading?

The gas test proves the preparation worked; it does not replace it. Before the tester arrives, the space has to be cut off, in order, from everything that could refill it:

1. Isolate by blinding, not by valves. A closed valve is a decision someone can reverse; a spade or a physical disconnect with an air gap is a fact. Every line in and out of the space is blinded or disconnected, and the permit lists each one by number — “all lines isolated” is not a record, it is a hope.
2. Drain, flush and purge whatever the space held. Product left in a dead leg or soaked into scale keeps giving off vapour long after the space looks empty, which is why a space that tested clean in the morning can fail at noon.
3. Ventilate with a purpose. Blowing air at the manway moves the reading, not the hazard. Effective ventilation puts the delivery point deep in the space, pulls from the far end, and keeps running for the whole entry — and if it stops, everyone comes out, because the ventilation was a condition of the permit, not a comfort.
4. Isolate mechanical and electrical energy too. Agitators, screw conveyors, and anything that can move is locked out before entry; atmosphere is the most famous confined space killer, but it is not the only one.

During the entry, monitoring does not stop. A pre-entry test is a snapshot of a space that breathes — disturbed sludge, warming walls and the work itself all change the air. Continuous monitors with alarms, worn in the breathing zone or placed at the work level, are standard practice at serious operators; where readings are taken periodically instead, the permit states the interval and someone actually keeps it.

What must a confined space permit record?

A confined space permit is only as good as what it forces someone to write down, not what it lets them assume. At minimum, it has to record:

The space and the exact task. The isolations and how they were proved — blanks and blinds listed, not just “isolated”. Every gas reading with the time it was taken and the name of the person who took it. Ventilation arrangements, and whether they run continuously. Access and lighting, both rated for the atmosphere. Communication method. The rescue plan and who is standing by. The names of entrant, attendant and supervisor. And the expiry — the moment the authorisation dies.

If a confined space entry involves welding or cutting, a hot work permit is raised in addition, and the two are cross-referenced. Neither replaces the other.

Country instruments

The confined space rules described above are close to universal in substance; what differs by country is which instrument makes them binding.

The failures that recur

The failures that keep recurring are not exotic — they are the same handful of shortcuts, repeated on different days. The reading taken at the manway instead of inside. The second person who follows the first without being on the permit. Ventilation switched off because it was noisy. An attendant pulled away for ten minutes to do something else. A permit written for one shift and quietly used for the next. Each one is small, and each one has killed people. None of them is prevented by the form of the record, but the form decides how fast the pattern is seen: a drawer of paper hides repetition that a searchable record shows in one query.

Standards cited

• 29 CFR 1910.146 — OSHA, United States · Permit-required confined spaces · Entry permit required before any entry into a permit space · In force
• NOM-033-STPS-2015 — STPS, Mexico · Condiciones de seguridad para realizar trabajos en espacios confinados · Written authorisation required for every confined-space entry · In force
• GB 30871-2022 — SAMR / SAC, China · 危险化学品企业特殊作业安全规范 · Eight special operations in hazardous chemical enterprises; a permit is mandatory for each · In force

Frequently asked questions