Why a Stair Pressurization System Fails Quietly

Why this matters

A pressurized stair is the only life safety system in a building that fails in both directions. Too little pressure and smoke enters the one path everybody is using. Too much and the door will not open, which for a person carrying a child or using a walker is the same outcome as a locked door. Those two bounds sit close together, the system has almost no headroom between them, and every ordinary thing that happens in a building over five years - a new rated wall, a worn door seal, a propped door, a painted damper - moves the number without anybody noticing. This card gives you one gate that checks both bounds at once and runs it against two stairs in the same building.

The gate, with its convention attached

Measure the differential pressure across the stair door, stair side minus building side, in inches of water column, with the doors in the position the design states, and check it against the lower bound and the upper bound in the same reading.

Every clause is a basis and it travels with the number.

• Across the stair door, not across the vestibule, not across the shaft wall. Different barrier, different number.
• Stair side minus building side, so a positive value means the stair is higher and air flows out of the stair into the floor, which is the direction that keeps smoke out.
• Inches of water column, and 1 in. w.c. equals 5.202 lbf per square foot of face pressure.
• Doors in the stated position. All doors closed is one criterion. The design number of doors open, which represents people actually evacuating, is a different criterion with a different acceptance number. A pressure reported without its door condition is not a measurement.

The lower bound comes from NFPA 92, in the edition your authority having jurisdiction has adopted and amended, whose minimum design pressure difference depends on ceiling height and on whether the building is sprinklered; in a sprinklered building the minimum with all doors closed is commonly 0.05 in. w.c., and the nonsprinklered values are higher and climb with ceiling height. The AHJ is a named role with authority to interpret and enforce the adopted edition, not a synonym for whoever inspects.

The upper bound is not a pressure at all. It is a force. NFPA 101, in the adopted edition, caps the force required to set a side-hinged swinging door in motion, commonly stated as not more than 30 lbf. Pressure only matters to the upper bound through the door, so you convert.

Converting pressure to the force a person actually feels

The differential acts over the whole leaf and pushes at its center. The person pulls at the handle, near the latch edge, so the handle has mechanical advantage about the hinge. For a leaf of width W with the handle a distance d from the latch edge, and uniform pressure across the leaf, the handle force from pressure alone is the total face force times W divided by 2 times (W minus d).

For a 36 in by 84 in leaf with the handle 3 in from the latch edge, that ratio is 36 divided by 66, about 0.545, and the leaf area is 21 sq ft. That 0.545 holds constant a 36 in leaf, a 3 in handle setback, a hinge at the opposite edge, and pressure uniform across the face. A 42 in leaf changes both the area and the ratio, so recompute rather than carrying 0.545 across.

Run the two ends of the range through it.

• At 0.10 in. w.c.: 0.52 psf times 21 sq ft is 10.9 lbf on the leaf, times 0.545 is 5.9 lbf at the handle. Add a door closer sized for the leaf, which typically contributes on the order of 8 to 12 lbf at the handle, and the total lands in the mid teens. Comfortable.
• At 0.35 in. w.c.: 1.82 psf times 21 sq ft is 38.2 lbf on the leaf, times 0.545 is 20.8 lbf at the handle. Add the same closer at 10 lbf and the total is about 31 lbf, over the 30 lbf cap.

That arithmetic is why the usable band on a common door is roughly 0.05 in. w.c. at the bottom and something in the low thirties of hundredths at the top, and why "turn the fan up" is not available as a fix. The closer eats a third of the budget before the fan does anything.

Why it goes quiet

The stair is a plenum. Everything that touches its air balance changes the number everywhere, not locally.

        roof relief damper
              |
   fan ---> [ top injection ]
              |
    +---------+---------+  stair shaft
    |  floor 11 door    |---> leaks to floor
    |  floor 10 door    |---> leaks to floor
    |     ...           |
    |  floor 2 door     |---> leaks to floor
    |  discharge door   |---> leaks to outside
    +-------------------+

  Pressure at any door is what the fan delivers minus
  everything that has already leaked out above it.
  Close one leak and every door below it rises.

Four ordinary changes move the number, and only one of them looks like fire protection work.

• A propped door. One stair door wedged open on any floor collapses the differential on every floor. This is the most common condition and the shortest lived, which is why it is almost never present during a test.
• A closed leakage path. The design counts on air bleeding out of the stair, through the floors, and out of the building. A tenant fit-out that adds a rated corridor wall removes part of that path, and the pressure at that floor goes UP. This is the one that runs against instinct: reducing leakage does not improve the system, it pushes the door force toward the cap.
• A changed relief path. A barometric or modulating relief damper stuck, painted, or its static sensing tube crimped takes away the system's ability to hold the top end, and the failure shows up as a door nobody can open.
• Weather. Stack effect adds to the differential at one end of the building and subtracts at the other, and which end depends on the sign of the indoor-outdoor temperature difference and where the neutral plane sits. NFPA 92, in the adopted edition, requires seasonal extremes to be considered, which is exactly why a system commissioned in October at a 10 F indoor-outdoor difference has not been tested at the 60 F difference it will see in January.

One gate, two stairs, opposite outcomes

Twelve-story sprinklered office building, two stairs, top injection on both, annual test under the adopted building code. All doors closed, force gauge at the handle, differential read through the listed test port in each door rather than by drilling anything. The stair was cleared and posted at each end before any door was opened, and no door was wedged at any point, because wedging one dumps the system for the entire building and is an impairment, not a testing technique. The fan room starter was locked and tagged before any guard came off, under 29 CFR 1910.147. Roof work at the relief damper was done tied off under 29 CFR 1910.28 in general industry, or 1926 Subpart M on a construction site.

Stair A, north. Floor 2 reads 0.06 in. w.c., stair minus building, all doors closed. Floor 11 reads 0.09. Both clear the 0.05 lower bound. Convert the floor 11 reading: 0.09 times 5.202 is 0.468 psf, times 21 sq ft is 9.8 lbf on the leaf, times 0.545 is 5.4 lbf at the handle. The measured closer contribution, taken with the fan off, is 11 lbf. Total measured force at the handle with the fan running: 17 lbf, against the 30 lbf cap. Stair A passes both bounds at both ends.

Stair B, south. Floor 6 reads 0.34 in. w.c. Convert: 1.769 psf, times 21 sq ft is 37.1 lbf on the leaf, times 0.545 is 20.2 lbf at the handle, plus the same 11 lbf closer, total about 31 lbf. Over the cap. Floor 2 of the same stair reads 0.04 in. w.c., below the 0.05 lower bound.

So Stair B fails the upper bound six floors up and the lower bound two floors up, on the same test, with the same fan. Both readings have one cause each and they are unrelated. Floor 6 got a rated corridor wall in a fit-out four months earlier that closed off part of the leakage path, so pressure there rose. Floor 2 is starved because the ground-level stair discharge door has a worn bottom seal with a visible gap, bleeding the bottom of the shaft to outside.

Correction, printed. The floor 6 fix is not to reduce fan output, because that would push floor 2 further below the lower bound. It is to restore relief: the roof relief damper was found with its linkage bound, and freeing it plus re-verifying the modulating control's sensing line brings the top end down without touching the bottom. The floor 2 fix is the discharge door seal. Both are re-tested at the same time, because either one alone changes both ends.

What each reading already contained. The 11 lbf closer contribution was measured with the fan off and then carried into the running-fan total as an addition, which is correct here because the fan-off reading contains no pressure component. Had the closer force been taken from a manufacturer's setting sheet instead, it would already contain a nominal door weight and no pressure at all, and it would still be an addition, not a re-basing. The 0.34 in. w.c. reading, by contrast, already contains the effect of the new rated wall, so the fix is not modeled as subtracting the wall, it is modeled as restoring the relief path the wall stole.

Sibling-rule check. Every pressure above carries its barrier, its direction, and its door position; every force carries its measurement point at the handle. No door was wedged or held, so the egress-path card's obstruction rule and the locked-door card's one-motion gate both hold throughout the test, and the 30 lbf cap used here is the same cap that card names. The smoke control card owns makeup air and interface height and is not re-derived here. The bound out-of-service period while the relief linkage was freed carries a start, an owner, and an end, which the records card requires.

What flips the answer

Multiple injection changes the diagnosis, not the gate. A stair fed at several levels distributes pressure far more evenly, so a single low floor reading points at a local leak rather than at shaft losses. Vestibules change the barrier: with a vestibule, the design may state the differential across the vestibule door, the stair door, or both, and reading the wrong one gives a number that is right and useless. And a stair whose discharge is through the building rather than directly outside is coupled to the lobby's own pressure, so lobby doors propped open during a delivery move the stair reading, which is a maintenance condition nobody thinks of as one.

References

• NFPA 92, Standard for Smoke Control Systems, in the edition adopted and amended by your authority having jurisdiction, for minimum design pressure differences and seasonal test conditions
• NFPA 101 Life Safety Code, in the adopted edition, for the maximum force to set a door in motion
• The adopted building code chapter on smoke control systems, for acceptance and periodic testing intervals
• 29 CFR 1910.147 for lockout and tagout of fan equipment; 29 CFR 1910.28 for fall protection in general industry, 29 CFR 1926 Subpart M in construction
• See related: What a Smoke Control System Is Trying to Achieve; Why a Locked Exit Door Is the Oldest Fatal Mistake