The Dizzy Tech After the Nitrogen Test

If someone is dizzy in a small space with nitrogen running

Read this part first if it is happening now.

• The person gets out, immediately, to open air. Not after finishing the braze, not after closing the cylinder. Dizziness, headache, fast breathing, clumsy hands or confusion while nitrogen is flowing in a closet means the air may be short of oxygen.
• Close the cylinder valve only if you can reach it from outside the space, from the doorway, without putting your head in. Otherwise leave it.
• If the person is down or cannot get out, do not go in after them. Call 911, prop every door open, and blow fresh air into the space with a fan from the doorway. Most deaths in oxygen-deficient spaces are people who went in to help.
• Once out, the person sits in fresh air and is watched. Anyone who lost consciousness, even briefly, or who stays confused or short of breath after a few minutes outside, goes to emergency care.

Nitrogen gives no warning. It has no smell or color, and the urge to breathe is driven by carbon dioxide building up in the blood, not by oxygen running low. A person breathing nitrogen-diluted air keeps clearing carbon dioxide normally, so they do not feel short of air. Judgment and coordination go first, which is exactly what someone needs to recognize the problem and leave.

The first event, and the right diagnosis

A ten-person HVAC shop was replacing an evaporator coil in a hall closet: 3 feet by 4 feet by 8 feet high, about 96 cubic feet of air, holding a gas furnace and the coil. After brazing, the tech, Aaron, set up a nitrogen sweep to push moisture out of the new coil and line set before the pressure test. Nitrogen flowed in at the outdoor unit and out of the open service port at the coil, inside the closet. The regulator had no flowmeter, so the flow was whatever the gauge setting gave, which was a lot more than a brazing purge. He pulled the door to within a few inches to keep the furnace noise out of the house and spent about half an hour in the closet on the condensate line while the sweep ran.

He came out with a headache, feeling light and slow, and told the homeowner he needed a minute. Ten minutes in the yard and he felt normal. He reported it at the end of the day, which is the reason this case exists.

The owner worked out what had happened, and got it right. Normal air is 20.9 percent oxygen. OSHA treats an atmosphere below 19.5 percent as oxygen-deficient (that is the definition in 29 CFR 1910.134 for respirators and 1910.146 for permit spaces). To pull 20.9 down to 19.5 takes very little: 1 minus 19.5 divided by 20.9 is about 0.067, so roughly 7 percent of the closet's air replaced by nitrogen, about 6 and a half cubic feet out of 96.

The sweep had moved far more than that. The owner could not know the actual flow, so he worked an illustrative one: at 40 standard cubic feet per hour, which an unmetered regulator can easily deliver, half an hour puts 20 cubic feet of nitrogen into the closet. If the closet were well mixed and the displaced air pushed out the door gap, oxygen falls as 20.9 times e to the minus 20 over 96, which is about 17 percent. That is not a measurement. It is the right order of magnitude for a headache and slowed thinking, and it says the gap between "fine" and "deficient" was a few minutes of sweep.

So the diagnosis was correct: nitrogen discharged into a small, nearly closed space displaced enough oxygen to impair the tech. The shop wrote two rules the next morning:

1. Nitrogen discharges outdoors. Any sweep, purge or test release runs through a hose to outside, never out of a port into the space.
2. The door stays open for any nitrogen work in a closet.

Both rules were right. Two months later it nearly happened again anyway.

The second event, with both rules followed

EVENT 1 (hall closet)          EVENT 2 (condo closet)
+------------------+           +------------------+
| AH      [port]   |           | AH  [port]--hose-|--> out
|          N2 out  |           |                  |  window
|          HERE    |           | [cyl] reg leak   |
| tech kneeling    |           | tech kneeling    |
+--[door 4 in]-----+           +--[door open]-----+
   cylinder outside at            laundry alcove:
   the outdoor unit               no window, no fan
AH = air handler and coil

A different tech, Lena, was pressure-testing a new air handler in a condo utility closet about the same size. The condo had no outdoor unit at ground level she could reach, so the nitrogen cylinder was strapped upright inside the closet beside the air handler. She followed the rules exactly: the door was wide open, and the test charge would be vented through a hose run across the laundry alcove and out a window.

The closet opened into a windowless laundry alcove, and the alcove opened into a hallway. The door was open onto air that did not move.

About 25 minutes into watching the test gauge, she noticed she was breathing fast and had a dull headache. She also noticed a faint hiss she had been hearing for a while without registering it. She walked out to the hallway and then outside, which the shop's training had taught her to do. Back in, from the doorway, she closed the cylinder valve and the hiss stopped.

The leak was at the regulator's inlet connection to the cylinder: a damaged sealing surface on the regulator nut, which had been fine when it was checked months before and had been dropped since. It was releasing nitrogen continuously into the closet, upstream of everything the vent hose controlled.

Why being right was not enough

The first diagnosis and both rules were aimed at the planned release: the nitrogen the tech meant to let out. That release is visible to the tech in the sense that he chooses it, so a rule about where it goes works.

The second event was an unplanned release: a leak at a fitting that nobody was watching. The vent hose did nothing for it, because the leak was on the cylinder side of the hose. The open door did less than the rule assumed, because an open door onto a dead-air alcove is not ventilation. A door lets air move; it does not make air move.

Three things the first fix silently assumed turned out to be false:

• That the cylinder would be outside. In the first event it was at the outdoor unit. In a condo, an apartment or an attic install, it often comes inside with the tech.
• That the only nitrogen in the space would be the nitrogen being discharged. A regulator connection, a hose barb or a cracked hose can release continuously at a rate too small to hear over a running blower, and over 25 minutes in 96 cubic feet a small leak is the same arithmetic as the first event.
• That an open door means fresh air. It does when it opens onto a large, ventilated room. It does not when the next room is small and still.

The only control that catches all three is one that measures the air rather than assumes it. A person cannot detect oxygen deficiency in themselves reliably; Lena noticed hers, Aaron noticed his late, and the shop had no reason to think the next tech would notice at all.

The rule that followed

The owner replaced the two rules with one procedure for any nitrogen work (brazing purge, sweep, standing pressure test) in a space smaller than a typical bedroom, which the shop defined as under about 1,000 cubic feet.

• Monitor, at the breathing zone, for the whole job. Every tech carries a personal gas monitor with an oxygen sensor, clipped near the collar, bump-tested before the day's first use, with its low-oxygen alarm at 19.5 percent. If it alarms, the tech leaves first and diagnoses from outside. A monitor clipped to a belt at knee height in a closet reads different air from what the tech breathes.
• The cylinder stays out of the space whenever there is anywhere else to put it, with the hose run in. When it cannot (an attic, a condo closet), it is leak-checked at the regulator connection with leak solution after hookup, not just when the regulator is new.
• Moving air, not just an open door. A fan blows fresh air into the space from a room that has its own supply, or from outside, for the whole job. The test release still goes outdoors through a hose.
• Metered flow. A flowmeter on the regulator for any purge or sweep, set to the low rate the job needs. An unmetered regulator is how the first event put 20 cubic feet into a closet.
• Somebody knows. A tech doing nitrogen work alone in a small space texts the office at start and at finish, and the office calls at a set interval in between. The shop set 20 minutes as its default.

None of this makes a utility closet a permit-required confined space. A closet you can walk into and out of is not one. But oxygen deficiency does not care what a space is called, and the controls a confined space uses for it (measure the atmosphere, ventilate, have someone outside who knows you are in there) are the ones that would have caught both events.

How the shop knew the new procedure worked

The owner did not wait for a third event to find out. On the next condo job with the cylinder inside, the tech logged the monitor's oxygen reading at the start, at the 15-minute mark and at the end of the test: 20.9, 20.9, 20.9, with the fan running and the regulator connection bubble-tested clean. On a later job the monitor's low-oxygen alarm never fired, but its record showed oxygen sagging to 20.4 near the end of a long brazing purge in an attic with the fan off by mistake, which is the kind of slow drift no person notices and every monitor does. The tech restarted the fan and the reading returned to 20.9 within a few minutes. That reading changed nothing about the job and justified the monitor in one afternoon.

What would change the answer

A larger, ventilated space. In a basement mechanical room of several thousand cubic feet with an open stair, the same leak that mattered in a closet is diluted many times over. The shop's 1,000-cubic-foot line exists so techs are not doing this arithmetic on site; above it, the monitor still goes on, and the fan is a judgment call.

A refrigerant release instead of nitrogen. Refrigerants are heavier than air and pool low, and hot surfaces can break them down into toxic products. That is a different hazard with its own card. See related: Refrigerant Decomposition and Oxygen Displacement.

References

• 29 CFR 1910.134(b), definition of an oxygen-deficient atmosphere (below 19.5 percent oxygen); 29 CFR 1910.146 for permit-required confined spaces
• Compressed Gas Association guidance on oxygen-deficient atmospheres and safe handling of compressed gas cylinders
• Equipment manufacturer instructions for nitrogen purge and pressure testing
• See related: Nitrogen Purging and the Air You Cannot Smell
• See related: How to Ventilate a Small Space You Are Working In