When the firestop fails, the crew is in the fire

Overhead cable tray passing through a purple fire-rated gypsum board wall assembly with red firestop material sealing the opening

A rated wall only protects people if openings through it stay protected. An unsealed penetration or unfinished joint lets smoke and hot gas leave the compartment while the wall still stands. Independent inspection and complete firestop help keep that boundary intact for occupants and responding crews.

Fire Examiners field team · Updated October 2026

A rated wall is a working boundary. People on the far side are supposed to have a corridor they can still use. The engine company is supposed to have a place to stretch a line, search, and fall back. Both depend on the same thing. The compartment has to stay a compartment.

An unsealed pipe, an empty sleeve, or a head-of-wall joint that was never finished does not wait for the gypsum to burn through. Smoke and hot gas move through the opening while the wall is still standing. The next room is already the fire. That is the building a crew walks into when the protection was never finished.

What they are breathing on the other side

People picture a fire death as contact with flame. The records do not.

NFPA’s reporter guide says most fire deaths are caused by smoke inhalation, not burns. Carbon monoxide is the most common toxic gas in that smoke, because it takes the place of oxygen in the blood. Hydrogen cyanide, from burning plastics, is the other gas that blocks the body’s use of oxygen. Superheated gas burns the airway. One breath of air that hot can kill.

The U.S. Fire Administration’s look at residential-building fatalities in 2017–2019 is a split, not a trend. Where a primary symptom was recorded, thermal burns and smoke inhalation together accounted for 89 percent. Burns and smoke combined were 49 percent. Smoke inhalation alone was 35 percent. Thermal burns alone were 6 percent. A symptom was recorded for 45 percent of the fatalities in that set.

Flame is real. It is not what crosses an open penetration first, and it is not what most of those fatalities were. The crew is in that air with a hoseline and an air pack. A barrier that already leaked has taken the tenable space they were counting on for the search.

How little an open hole buys

These are not death rates. They are counts of openings that were never closed, and of how fast an unproven opening stops being a barrier.

A 2008 New Zealand pilot looked at 11 buildings: hospital, student housing, residential, and office. Eight of the 11 had passive fire protection that would likely fail in a fire. Bad service penetrations were the most common miss, and much of that work was already covered up. Eleven buildings is an indication, not a national rate. On 11 Auckland records from late 2016, about half the penetrations were completely unsealed. The repeats were bare cables and pipes, sealant on one side of a floor only, and systems without the plasterboard depth the product needs.

When those unproven details were furnace-tested toward a nominal 60 minutes, the earliest insulation failure was 2 minutes. The earliest integrity failure was 19 minutes. Those are the earliest results in the set, not a claim that every open hole fails in two minutes. A related test, at a higher pressure difference, lost integrity in as little as 10 minutes through a gap of 2.6 mm. An annular space the listing does not allow is not a cosmetic miss. It is a path the fire can use inside the first half hour, while a listed seal is still supposed to be holding.

The nearest U.S. incident count is older, and it is not a study of listed ASTM E814 systems. In 1977 the National Bureau of Standards reviewed 84 fires in low-rise apartments around Washington, D.C., and only fires that had already left the unit of origin. Seventy-nine spread into another apartment. Partitions stacked with no firestopping at the floor or ceiling showed up in 24 of them, 29 percent. A pipe or duct through the protective skin of a wood-stud chase showed up in 20, 24 percent. “Firestopping” in that report means blocking a concealed draft, not a listed through-penetration. The pipe-and-duct item is the one that still matches the work: the wall was pierced and not closed, and the fire used the opening.

A UK review of investigations from 2003 to 2013 found the same pattern in concealed spaces, and the contrast. Of about 106 investigations, 34 involved fire moving through a hidden space. In 7, the wall-to-roof joint was not firestopped or the seal did not reach the roof. Cavity barriers were missing or badly fitted in 20 of the 34. In 3, the roof-void compartment was there, and it limited the spread. Where the barrier held, the fire stayed on its side of the line. Where it did not, the fire was already past the boundary the crew thought they had.

Yellow tape measure stretched across a rectangular floor slot opening containing metal conduits and mineral wool backing

What has to be in the opening

A through-penetration listing under ASTM E814 and ANSI/UL 1479 is the lab proof for one specific seal, in one wall or floor type. The sample sees a standard fire. It has to keep flame off the unexposed side, and it has to stay in the opening when a hose stream hits it afterward. The stream is a standard blow for impact, erosion, and cooling. It is not a replay of fireground tactics. It is there because a thin coating can look fine in the furnace and come apart the moment it is hit. The crew’s line is that hit.

The wall’s own ASTM E119 or ANSI/UL 263 rating does not cover the hole cut through it. Joints are a separate listing, ASTM E1966 and ANSI/UL 2079. The seal in the opening has to be the listed seal for that opening. Anything else is the condition in the studies above: a rated wall with a path through it.

The field check, before the ceiling goes on

Firestop Special Inspection does not repeat the furnace, and it does not put a hose on a finished wall. ASTM E2174, for penetrations, and ASTM E2393, for joints and perimeter fire barriers, check that the listed system is the one installed, the way the listing describes. The International Firestop Council’s summary of the practice is a sample. For each type of penetration firestop, witness 10 percent of the installations, or destructively test 2 percent. A failed installation is repaired. The practice does not allow a leftover deficiency.

The model code requires that inspection in high-rise buildings and in Risk Category III or IV. The 2021 International Building Code also requires it in Group R fire areas with an occupant load greater than 250. A state or city adopts the edition on its own schedule, so an AHJ on an older code may not have the newer trigger. Where the inspection applies, the owner or the owner’s agent hires the approved agency, and that agency is independent of the contractor doing the work. Section 110.3.6 keeps joints and penetrations in fire-resistance-rated assemblies, smoke barriers, and smoke partitions from being covered until they are inspected and approved.

That is the protection the listing was written to provide. The seal that stayed in place through the furnace and the hose stream is the seal in the wall. The compartment the crew is counting on is still a compartment.

If you need the inspection scope written for a project, call Fire Examiners at 727-537-0047 or use the contact form.