Why an Excavation Can Become a Hazardous Atmosphere

Why this matters

Nobody enters an excavation to test it. Testing is done from the surface, with a probe and sample line, before anyone is exposed, and if a reading comes back bad while people are in the cut they come out under 29 CFR 1926.651(k)(2) before anything else is decided.

The thing that surprises crews is that an open trench has its own atmospheric standard, separate from confined spaces, and its trigger is not depth. 1926.651(g)(1)(i) requires the atmosphere to be tested before employees enter excavations greater than 4 feet in depth where oxygen deficiency or a hazardous atmosphere exists or could reasonably be expected to exist, and it names the kind of reason it has in mind: excavations in landfill areas, or in areas where hazardous substances are stored nearby. The 4 feet is a floor on when the duty can attach. The trigger itself is a question about the neighbourhood and the history, and it is answerable before you mobilise.

An open cut is not a confined space, and that is a scope fact worth knowing

The scope provision of Subpart AA, the confined spaces in construction standard at 29 CFR 1926.1201, excludes construction work regulated by Subpart P. So an excavation, worked as an excavation, is governed by 1926.651(g), not by the permit-space program.

Where that fork bites is at the boundary. Digging the trench down to a sewer is Subpart P work. Opening that sewer and putting a person inside it, or into a manhole reached from the trench, is not - that is a confined space and it comes with a different program, a different set of duties and a different entry regime. Crews cross that line without a stop, because it feels like a continuation of the same job. It is not.

What actually fills a cut

An excavation is a hole with no natural ventilation at the bottom. At a given wind speed, the deeper and narrower the cut, the less surface air exchange reaches the bottom, which is why a 3-foot-wide, 9-foot-deep trench behaves nothing like a 15-foot-wide pit at the same depth. Plates over the road, a tarp left on overnight, or a still high-pressure morning all reduce it further.

Four families of source, and they are worth separating because the crew can only control two of them.

• What is below or beside the cut. A leaking gas service or main, a sanitary or storm line, contaminated fill, a former tank pit, a landfill, decomposing organic material.
• What is running at the edge. A generator, a compressor, a pump, a truck idling, a plate compactor. Every internal combustion engine is a carbon monoxide source discharging into a hole.
• What the crew brings down. A gasoline cut-off saw, a plate compactor working the bottom, torch work on copper, solvent cement on plastic pipe, a portable heater.
• Displacement. A gas that is not toxic can still push oxygen out. Below 19.5 percent oxygen the atmosphere is deficient by the standard's own definition in 1926.651(g)(1)(i).

Density matters and cuts both ways. Gasoline vapour, many solvent vapours, propane, hydrogen sulfide and carbon dioxide are heavier than air and settle into the bottom, so a reading taken at the lip does not represent where the person's head will be. Methane and hydrogen are lighter and will collect at the top or under a plate, so a reading at the bottom does not represent a pocket at the top. Carbon monoxide is close to air density and distributes fairly evenly, which is precisely why an engine fills the whole cut rather than pooling in one corner. Test at multiple levels, top, middle and bottom, on the way in.

The gate, and two jobs it answers differently

Ask, before mobilising: is there a reason from this site's history, its surroundings, or the planned work to expect a hazardous atmosphere in this cut?

Job A: an 8-foot water service trench in a 1970s subdivision. Open front lawn, no fuel storage on the block, no landfill, no sanitary line being opened, dewatering pump is electric off the house service. Trigger reviewed and not met. Record it either way, because "reviewed, not met, here is why" is a real entry and a blank is not.

Job B: a 7-foot sanitary lateral, 30 feet downgradient of a lot that held a filling station until the tanks came out in the 1990s. Trigger met on site history alone, before a shovel moves. That means, in order: test before entry from the surface, ventilate to control what is found, keep testing under 1926.651(g)(1)(iv) as often as necessary while controls are relied on, and have emergency rescue equipment readily available and attended when in use under 1926.651(g)(2)(i). The flammable-gas control number is in 1926.651(g)(1)(iii): adequate precaution such as ventilation to prevent exposure to an atmosphere containing a flammable gas in excess of 20 percent of the lower flammable limit. That is a ventilate-now threshold, not a leave-when-you-reach-it threshold.

Job A at 10:15, and the arithmetic that gets people hurt

The gate is not asked once. Job A cleared it at 06:40 and stopped clearing it at 10:15, when the crew brought a gasoline cut-off saw down to cut ductile iron in the bottom. The source changed from the neighbourhood to the toolbox, and nothing on the morning sheet covers that.

Say the personal monitor on the man in the cut reads 180 ppm carbon monoxide 45 minutes into the cutting. The exposure limit that applies to construction work through 1926.55 is 50 ppm as an 8-hour time-weighted average. Someone does the obvious sum:

• 180 ppm x 0.75 h = 135 ppm-hours
• 135 / 8 h = 16.9 ppm contribution to the 8-hour TWA
• 16.9 against 50: a comfortable pass

That arithmetic is correct and the conclusion is wrong, in the flattering direction, by a factor of 180 / 16.9, which is about 10.7. Three reasons, and they are all about the basis of the number rather than the sum:

• The bases do not match. 50 ppm is an 8-hour time-weighted average. 180 ppm is an instantaneous reading. Averaging an instantaneous reading against a shift-length basis is not a comparison, and the person in the cut is breathing 180, not 17.
• It is a rising series, not a sample. An engine running in a still cut fills it. There is no plateau to average against, and the next reading is higher than this one unless something changes.
• The removal duty is not conditioned on a TWA. 1926.651(k)(2) requires exposed employees to be removed where there is evidence of a hazardous atmosphere, until the necessary precautions have been taken. That is a stop rule, and it does not wait for arithmetic.

What actually happens. The saw stops. Everyone leaves the cut by the nearest ladder. The saw comes out of the trench, and the work is either done with a tool that does not run an engine in the hole, or the cut is mechanically ventilated with clean air drawn from a point well clear of any exhaust and the atmosphere is re-tested from the surface before re-entry, then monitored continuously while the work runs. If ventilation is the control, 1926.651(g)(1)(iv) requires the testing to continue as often as necessary to keep the atmosphere safe, and continuous monitoring on the person is the practical form of that.

The failure mode. Nobody in this story decided to take a risk. They ran an engine in a hole, which is a thing done on thousands of jobs, and the one person with a number in front of them reached for a limit whose basis they had not read. Carbon monoxide is odourless, so there is no cue except the instrument, and the first symptom in a working man is usually a headache he blames on the heat.

Two more Subpart P clauses that live here

Respiratory protection is not the first answer. 1926.651(g)(1)(ii) requires adequate precautions against atmospheres below 19.5 percent oxygen and other hazardous atmospheres, and names respiratory protection or ventilation under the relevant subparts. Ventilation removes the hazard; a respirator manages an exposure and brings a whole program with it - written procedures, medical evaluation, fit testing, training. In an open cut with a controllable source, moving or eliminating the source is almost always the cheaper and more reliable control, and it is the one that also protects the person who walks up without a respirator.

Bell-bottom pier holes get their own rule. 1926.651(g)(2)(ii) requires an employee entering a bell-bottom pier hole or similar deep and confined footing excavation to wear a harness with a lifeline securely attached, the lifeline separate from any line used to handle materials, and individually attended at all times while the employee is in the excavation. Field-service shops meet this on sign, pole and light-standard footings and rarely recognise it as a named case in the standard.

How to verify

1. Answer the gate in writing before mobilising, from site history rather than from the look of the ground. The two questions that find most of it: what was on this parcel and the ones next to it before, and what utilities run under and beside the alignment.
2. Re-ask the gate whenever the source list changes - a new engine at the edge, a tool going down, a line being opened, plates going over the cut, a tarp going on overnight.
3. Test from the surface, at three levels, before any entry, allowing the instrument's stated response time for the length of sample tubing in use. Never lower a person to place the probe.
4. Read every limit's basis before you compare anything to it. Write the basis next to the number on the sheet: 8-hour TWA, ceiling, percent of lower flammable limit, percent oxygen. A number without its basis cannot be compared to a reading.
5. Confirm rescue capability exists before the trigger is met, not after. 1926.651(g)(2)(i) requires the equipment readily available and attended when in use, and "readily available" is measured in the minutes an unconscious person has, not in whether it is somewhere on the truck.
6. Draw the Subpart P to Subpart AA line on the plan, and name the moment the job crosses it, so the crew meets a decision rather than a doorway.

References

• 29 CFR 1926.651(g), hazardous atmospheres in excavations, including the testing trigger and 19.5 percent oxygen at (g)(1)(i), precautions at (g)(1)(ii), the 20 percent of lower flammable limit ventilation threshold at (g)(1)(iii), continued testing at (g)(1)(iv), and emergency rescue equipment and bell-bottom pier holes at (g)(2)
• 29 CFR 1926.651(k)(2), removal of exposed employees where evidence of a hazardous atmosphere or other hazardous condition exists
• 29 CFR 1926.1201, scope of Subpart AA, confined spaces in construction, which excludes construction work regulated by Subpart P
• 29 CFR 1926.55, gases, vapors, fumes, dusts and mists, the route by which exposure limits including the carbon monoxide 8-hour time-weighted average apply to construction work
• See related: Testing the Atmosphere Before Anyone Enters a Wet Well or Manhole for the testing method itself; Confined Space Entry (OSHA 1910.146) and universal-why-a-sewer-is-a-confined-space-and-what-that-changes for the other side of the scope line