Why a Trap Seal Is the Only Barrier and What It Is Made Of

Why this matters

A building has a handful of things standing between the sewer and the air people breathe, and only one of them is made of water. That one can fail with nothing broken, nothing loose, nothing corroded and nothing to photograph, and it can put itself back before you arrive. Every other barrier in the list leaves evidence. So this article does two things: it names the five separations that are real, and it names the seven things technicians routinely inspect that were never holding anything back. Finding all seven intact and reporting no fault found is how a call with a real cause gets closed empty.

What counts as a barrier here

A barrier is a physical separation that a gas cannot cross while it is intact. That is a narrow definition on purpose. A device that dilutes, that admits air one way, that catches hair, or that keeps a smell from reaching one particular room is not a barrier, even when it is genuinely useful and even when its absence causes complaints.

Sewer gas is not one gas. It is a mixture whose hazardous components include hydrogen sulfide and methane, and its concentration in the piping is not something you control. You control whether it can get out. So the question on every smell call is not "is there gas in the pipe" - there is - it is "which separation is open."

The complete inventory of real barriers

There are five, and they are the only five in an ordinary building.

Read down the last column. Four of the five announce themselves. The trap seal is the exception, and that is the whole reason a smell complaint is hard: the only barrier that can be gone at 6 p.m. and present at 9 a.m. is the one you cannot inspect after the fact.

What is not a barrier, and why people keep counting it

• The strainer or floor grate. It stops hair and debris. Air moves through it as if it were not there.
• The fixture body. A sink, a shower base and a floor drain body are all open to the pipe by design. The porcelain is not holding anything back.
• The shape of the trap. A P-trap with no water in it is a piece of pipe with two bends. People say "there is a trap there" as if the fitting were the barrier. The water is the barrier and the fitting only holds it.
• An air admittance valve. It is a one-way check that lets air in and is designed not to let air out. It is a vent, not a seal, and it is the wrong device to reach for on a positive-pressure problem, which the base-of-stack article covers.
• A trap primer. It is a supply device that keeps a seal topped up. It maintains a barrier; it is not one, and a primer on a dry line is doing nothing at all.
• Terminal placement rules. Keeping a vent terminal away from openable windows and intakes is a dilution and placement control. It reduces what a person notices outdoors. It separates nothing.
• A closed door. Obvious written down, routinely relied on in practice when someone says the smell is only in the back room.

The geometry that sets the number

The seal depth is the vertical distance from the dip, the low point of the U, up to the crown weir, which is the top of the outlet leg where water spills over toward the drain. That vertical dimension is the entire specification. Trap material, trap body size, and how much total water the casting holds do not change it: a large-bodied trap and a slim one with the same dip-to-weir height are the same barrier.

Model codes commonly require a minimum of two inches of seal depth, as adopted and amended by your jurisdiction. Deeper seals exist and are used where evaporation or pressure is expected to be harsh, and they help, but the direction of that help is worth stating precisely: depth buys time against slow loss and margin against a transient. It does not change the mechanism by which a seal is lost, which the seal-loss article owns.

Reading a seal as a pressure device

Two inches of water does not mean the trap holds two inches of water column of differential. The static balance depends on the trap's leg geometry, and a real event is a transient with momentum behind it, which moves more water than a static balance predicts and can carry part of the seal away with the flow rather than merely displacing it.

What the trade designs to is a system pressure criterion: hold pressure inside the drainage piping within about plus or minus 1 inch of water column, stated against a nominal two-inch seal. Note the condition attached to that number. It was written for a full two-inch seal, so it does not transfer to a trap you have measured at less than that. A short seal is outside the condition the criterion was derived under, and the correct response is to restore the seal and then measure, not to scale the criterion down and keep going.

Worked example: the seal you have against the pressure the room makes

A mechanical room floor drain, intermittent sewer odor, worse when the exhaust fan runs.

Dip the trap before anything else. Use a marked probe or a length of stiff tube on a rule. Do not put your face over the opening to smell it: hydrogen sulfide deadens the sense of smell at concentrations well below the ones that incapacitate, so your nose reports "less" exactly when the exposure is worse, and the only reliable read is a calibrated instrument. Work from the side, upwind of the opening if there is any air movement.

• Nominal depth from the drawing: 2.0 inches
• Measured depth: 0.9 inches
• Shortfall: 1.1 inches

Apply the criterion's condition, do not scale it. The plus or minus 1 inch criterion is written for a nominal two-inch seal. At 0.9 inches this trap is outside that condition, so any pressure comparison made now describes a trap nobody designed. That is finding number one and it goes on the ticket before any pressure work.

Then measure the room. Digital manometer, 10 inches of water column full scale, specified at plus or minus 1 percent of full scale. Write out what that means before using it: 1 percent of 10 is 0.10 inches of water column, the basis is full scale so it does NOT shrink as the reading shrinks, and its character is a fixed systematic offset, the same sign on every reading this instrument takes today.

• Fan off, room referenced to the drain: 0.05 inches of water column negative. Against a fixed bound of 0.10, that single reading is indistinguishable from zero and must not be quoted alone.
• Fan on: 0.55 inches of water column negative.

Take the difference, and say why it is the usable number. 0.55 minus 0.05 is 0.50 inches of water column. Because the 0.10 is a fixed offset from one instrument, it appears in both readings with the same sign and cancels in the subtraction. What survives is only the instrument's residual nonlinearity and repeatability, a small fraction of the 0.50, rather than the full 0.10. So the honest statement is: the fan swings this room 0.50 inches of water column relative to the drain, and that number is trustworthy in a way neither endpoint is.

Check direction before combining anything. A room held negative relative to the drain means the drain side is relatively positive and pushes the seal toward the room. A back-pressure event at the base of the stack pushes the seal the same way. Two effects only add if they push the seal in the same direction, and here they do.

Combine as bounds, not as an interval. Say a separate peak measurement during a discharge test gives 0.70 inches of water column at this drain. These are worst-case bounds on two independent contributions, and worst-case bounds add linearly and are reported as bounds. So: total displacing pressure <= 1.20 inches of water column. One inequality, no plus or minus, because a plus or minus would claim knowledge of a distribution nobody measured.

Land it. 1.20 exceeds the plus or minus 1 inch criterion, and neither contribution exceeds it alone. That is the answer to why the complaint is intermittent: it needs the fan and a discharge at the same time. The fix list is now specific - restore the seal to full depth and keep it there, then either reduce the room's depressurization or relieve the stack event, because removing either one drops the total under the criterion.

The failure mode. The common wrong call is to quote the fan-off reading of 0.05 as proof the room is neutral. It is not proof of anything: the instrument's own fixed bound is twice that reading. A tech who reports "room pressure normal" from a single reading inside the instrument's noise has produced a sentence a second tech will trust and should not.

How to verify you measured the seal and not the trap

Three checks, each of which catches a different wrong reading.

Check you hit the dip and not a fitting shoulder. Sound the trap in two spots a quarter turn apart and take the deeper of the two. A probe that catches on a shoulder reads a shallow seal on a healthy trap, which sends you chasing a loss that never happened.

Check the trap is the one serving the complaint. In a room with several floor drains, mark each grate and run water into one at a time, watching for the level to rise in the one you marked. Buildings routinely have a drain that turns out to be tied to a different branch than the drawings show, and an abandoned drain with no primer and no traffic is the most common single source of a room odor.

Check the barrier list rather than the trap alone. Walk the five real barriers in the table above for that room: piping and joints visible, every cleanout plug present and tight, closet seals where there are closets, every abandoned stub capped. Photograph each one as you clear it. If all five are intact and the seal is full, the fault is upstream of this room, and the smell-complaint article picks up from there.

If any part of that walk would put you into a manhole, a lift station wet well or a large interceptor, stop: those are permit-required confined spaces and entry requires a permit, atmospheric testing, ventilation, an attendant and a rescue plan under 29 CFR 1910.146 in general industry or 29 CFR 1926 Subpart AA in construction. Brief whoever is standing with you that nobody follows anybody in, because untrained rescuer entry is how a single casualty becomes several.

References

• The plumbing code as adopted and amended by your local jurisdiction (the IPC or UPC family), which sets minimum trap seal depth, cleanout requirements and vent terminal placement, and binds through the adopting ordinance and the permit rather than on its own
• 29 CFR 1910.146 (general industry) and 29 CFR 1926 Subpart AA (construction), permit-required confined spaces, for any sewer manhole, wet well or interceptor
• Manufacturer documentation for the manometer in use, for the accuracy basis (percent of reading against percent of full scale) quoted in the worked example
• See related: What a Drainage System Is Actually Doing With Air; How a Trap Seal Is Lost and How to Tell Which Way It Went; What a Drain Trap Is Doing and Why It Matters