Why a Sewer Is a Confined Space and What That Changes

Why this matters

The thing that kills people in sewers is the air, not the water. And the specific cruelty of sewer air is that the human senses fail in the exact order that gets someone killed: the gas you can smell at harmless concentrations stops registering right around the concentration that will drop you, the two gases that displace your oxygen have no smell at all, and oxygen deficiency itself gives no warning of any kind before the confusion starts. A technician standing at an open manhole feeling fine is not evidence about the atmosphere four feet down. This card is about why every field substitute for an instrument and a permit is already on a list of things that have killed people, and what a shop has to have in the truck instead.

The classification, stated once

A space is confined when it is large enough to enter and work in, has limited means of entry or exit, and is not designed for continuous occupancy. A manhole, a wet well, a lift station dry pit, a large interceptor and a septic tank all meet all three without argument.

It becomes permit-required when it also contains or could contain a hazardous atmosphere, holds a material that could engulf an entrant, has an internal shape that could trap or asphyxiate, or carries any other recognized serious hazard. A sewer meets the first permanently, because it manufactures its own hazardous atmosphere out of what is flowing through it. The general industry standard is 29 CFR 1910.146; on construction work it is 29 CFR 1926 Subpart AA, and they are not identical documents. Name which Part the job falls under before the permit is written. Two sibling cards cover general classification and what a permit does to a working day; this one covers why the sewer is the hard case.

Four atmospheric hazards, and what each one warns you with

Two of those four have no warning property whatsoever, and the one that has the strongest warning property loses it fastest. That is the whole problem in one table.

The sense that fails at exactly the wrong concentration

Hydrogen sulfide has an odour threshold in the parts-per-billion range. You will smell it standing over an open cleanout at concentrations that are not going to hurt you, which is why the smell is such a familiar part of the work and so easy to treat as background.

Then, at concentrations on the order of 100 ppm and above, it paralyses the olfactory nerve. Not gradually and not partially. Within a minute or two of exposure at that level, the smell goes away. NIOSH puts the IDLH, the concentration immediately dangerous to life and health, at 100 ppm - the same order of magnitude.

Put the two facts in one sentence and you have the field rule: the smell getting weaker as you work is a symptom of rising concentration, not falling. Every technician's instinct reads it the other way, and that instinct is the mechanism by which people walk down a ladder into a lethal atmosphere feeling reassured. There is no version of experience that fixes this. It is nerve physiology, and it works the same on a thirty-year veteran.

For exposure limits rather than escape limits, OSHA's general industry ceiling for hydrogen sulfide sits in 29 CFR 1910.1000 Table Z-2 and the construction limit is set separately under 29 CFR 1926.55. Both are lower than the IDLH, and neither is a permit-entry criterion; the permit criteria are in 1910.146 itself.

The negative space: what does not count as a control

This is the useful half of the card, because every item below is something working technicians actually do, and each one is excluded for a specific reason rather than out of caution.

• Your nose. Excluded by the paragraph above. It works at concentrations that do not matter and stops at the ones that do.
• A reading taken at the open hatch. Excluded because gases stratify. Hydrogen sulfide has a specific gravity near 1.19 and settles low; methane is around 0.55 and collects high, under the lid. A single reading at the opening can miss both, in opposite directions.
• Leaving the lid off for a while. Excluded because passive airing does not reliably turn over the air in a deep space, and because the sewer keeps making more the entire time. Natural ventilation is not a control; forced ventilation with the discharge routed away from the opening and from anyone working, run continuously during the entry, is.
• Going in quick. Excluded because IDLH is defined by escape capability, not by dose. Above it, the question is not how long you can afford to stay, it is whether you could get yourself out if something went wrong. A short exposure is still a lethal one.
• Holding your breath. Excluded because at high concentrations hydrogen sulfide causes near-instant collapse, and because the entrant does not know the concentration before entering, which was the point of the instrument they skipped.
• Someone standing at the top watching. Excluded on its own. An attendant is required, and an attendant without non-entry retrieval rigged before entry, and without a rescue service that has been contacted and confirmed, is a witness rather than a rescuer.
• A rope tied around a waist. Excluded because a body cannot be lifted vertically past a shoulder-width opening on a waist tie without jamming. Retrieval means a full-body harness with a dorsal attachment and a mechanical device positioned over the opening.
• A single-gas monitor. Excluded because the table above has four rows. An instrument that reads hydrogen sulfide alone says nothing about an oxygen deficiency that is already present.

The arithmetic nobody wants to do

NIOSH has reported for decades that a majority of confined space fatalities are would-be rescuers, commonly cited at around 60 percent. Read that as a multiplier rather than a statistic: one entrant goes down and, on average, more than one additional person follows them in. The mechanism is not ignorance, it is that the atmosphere gave the rescuer no warning either, and the rescuer had less than a second to decide.

So the rule that follows is uncomfortable and it is absolute: nobody enters to retrieve. Ever. The attendant's job during an emergency is to summon rescue, operate the retrieval device from outside, and physically prevent unauthorized entry, including by co-workers who arrive after the fact. If your written program does not say who is empowered to stop a colleague from going in, it has a hole in exactly the place the fatality data points at.

Worked example: one wet well, three depths

A routine call to a below-grade lift station wet well: a pump is short-cycling and the customer wants a float checked. Nothing about the ticket says confined space, which is how most of these start.

The atmosphere is checked with a pump-driven multi-gas instrument, bump-tested that morning, sampling through a hose lowered from the surface. Nobody is over the opening while it runs.

Now apply the corrections the general section imposed, each on its own line.

Correction 1, the flammable reading is on the wrong calibration gas. A catalytic bead sensor calibrated on one hydrocarbon responds differently to another, and the raw number has to be multiplied by the response factor printed in that instrument's manual for that pairing. Say the manual gives 1.6 for methane on a pentane-calibrated sensor; use the factor from your own instrument's table, not this one.

• 7 percent LEL raw, times 1.6, equals 11.2 percent LEL corrected. Call it 11.

Correction 2, that corrected figure is a lower bound, not a value. A catalytic bead sensor needs oxygen to burn the sample, and at 17.4 percent oxygen it is under-reading. Hydrogen sulfide at 140 ppm also poisons that sensor family over time. Both errors run the same direction. So the honest statement carries one inequality sign and no interval: flammable at the bottom is at least 11 percent LEL, and 1910.146's hazardous-atmosphere criterion is 10 percent.

Correction 3, the hatch reading was never a reading of the space. 20.8 percent oxygen and 3 ppm at the opening described the air at the opening. Twelve feet down, oxygen is 17.4 percent, below the 19.5 percent floor, and hydrogen sulfide is 140 ppm, above the 100 ppm IDLH. Had the crew accepted the hatch reading, every one of the three hazards would have been missed.

What the nose would have said. At 3 ppm at the hatch, unmistakable rotten-egg smell. At 140 ppm at the bottom, gone within a minute or two of arriving. A technician who used smell as the instrument would have descended into an atmosphere reading worse on all three rows while experiencing it as improving.

The call. Three independent disqualifying conditions. No entry on the strength of ventilation and optimism. The float gets checked from the surface with a hooked pole if it can be reached, or the job becomes a scheduled permit entry with forced ventilation running continuously, a re-test after ventilation with the instrument's own response and sample-draw time allowed for at every depth, an attendant, retrieval rigged over the opening, and a confirmed rescue service. That is not the same job, it does not fit in the same slot, and pricing it as the original ticket is how a shop pressures its own crew into the shortcut.

What flips it. If the bottom sample had come back at 20.6 percent oxygen, 0 percent LEL corrected and 4 ppm, the space is still permit-required by classification, because a sewer's potential to contain a hazardous atmosphere does not go away between visits. What changes is whether the alternate-procedures path in 1910.146 is available for that entry, and that is a written determination made by a qualified person against the standard's own conditions, not a judgement call made at the hatch.

Confirming your setup is real, not nominal

Four checks a shop can run on itself this week, none requiring an entry:

1. Bump test versus calibration. A bump test proves the sensors respond; a calibration sets what they respond to. Both have intervals in the manual, and the bump test is the one that gets skipped. Check the instrument's log, not the sticker.
2. Wait time versus training. Response time is quoted for gas at the sensor; the sample still has to cross the hose. Ask the crew how long they hold at each depth. A fixed number smaller than the manual's is a training gap.
3. The retrieval device, rigged in the yard. A crew that has not physically set the tripod and davit over an opening in the last year will not do it correctly under pressure at a live station.
4. The rescue service phone call. Ask the number in the permit program what their response time is to that address and whether they perform sewer entries. A number nobody has dialled is not a rescue plan.

References

• 29 CFR 1910 Part 1910.146, permit-required confined spaces, general industry; 29 CFR 1926 Subpart AA for construction work
• 29 CFR 1910.1000 Table Z-2 for the general industry hydrogen sulfide ceiling; 29 CFR 1926.55 for the construction limit
• NIOSH Immediately Dangerous to Life or Health value for hydrogen sulfide, and NIOSH reporting on the proportion of confined space fatalities that are would-be rescuers
• Manufacturer documentation for the multi-gas instrument in use, for response factors, response time and calibration intervals
• See related: Confined Space Entry (OSHA 1910.146); What a Confined Space Permit Changes About Your Day; Testing the Atmosphere Before Anyone Enters a Wet Well or Manhole