How a Trap Seal Is Lost and How to Tell Which Way It Went

Why this matters

Every mechanism that empties a trap ends the same way, with an open pipe and a smell, so the symptom tells you almost nothing about the cause. What separates them is time: some losses accumulate with the clock whether or not anyone uses the building, and some happen per discharge event and cost nothing in between. Those two families need opposite repairs, and they are trivially easy to tell apart with one controlled test and a marked probe. Techs skip the test because the smell feels like enough evidence, then fit a trap primer on a fixture that is being siphoned, and the callback arrives the first busy morning.

First, the check that is not a mechanism

Before you diagnose a loss, confirm there was ever a seal. A trap that leaks through a slip joint, a trap someone removed during a remodel and never replaced, a floor drain that was installed with no trap at all, and an open stub from a demolished fixture are all common and none of them is a seal-loss mechanism. They are missing barriers, and the barrier inventory article covers how to walk them. Dip the trap. If the fitting is dry the day you refill it and dry again an hour later with nothing running, you have a leak, not a loss.

If the customer has poured a chemical drain cleaner into the line, treat every joint below that fixture as holding standing caustic. Before opening it, wear chemical splash goggles and gloves rated for the product on its safety data sheet, break the joint low and angled away from your body and face, never over your head, and put a catch pan under it. If you cannot identify what was poured, flush cold water from an upstream fixture until the line runs clear rather than opening blind. The next tech inherits whatever you leave in that line, so note the treatment on the ticket.

The two families

Time-driven losses accumulate with clock time and are indifferent to occupancy. A vacant unit loses seal at the same rate as a busy one. They are slow, they show up in units nobody uses, and they are quoted in inches per week.

Event-driven losses happen during a discharge somewhere in the system and cost nothing between events. They are fast, they show up in busy buildings at peak, and they are quoted in inches per event.

Those two units are not comparable. Inches per week and inches per event only meet after you multiply the second one by an event rate, and that multiplication is where most of the diagnostic value sits.

The six mechanisms and what each leaves behind

Direction is the discriminator that costs nothing to collect. Negative events pull air in through the seal, which is what a gurgle is. Positive events push water out of the seal toward the room, which is what a burp is. Same complaint, opposite pressure, opposite half of the system to fix.

The gate: refill, control the events, re-dip on a schedule

One test separates the families and it is worth doing properly rather than approximately.

1. Dip and record every trap on the affected branch, including floor drains people forget.
2. Refill each one to full depth.
3. Control the event variable. Either lock the branch out of use, or have the occupant log every discharge with a time.
4. Re-dip at 24 hours and again at 7 days.
5. Record your probe's resolution and never report a change smaller than it. A marked probe or stiff tube on a rule reads to about a sixteenth of an inch, so a change under 0.06 inches is not a reading, it is noise.

While dipping, keep your face out of the opening and work from the side. Hydrogen sulfide in sewer gas deadens the sense of smell well below the concentrations that incapacitate, so the nose reports less exactly when exposure is worse; if the space is confined or unventilated, a calibrated multi-gas instrument is the only honest read.

Evaporation from an undisturbed trap in a conditioned space commonly runs on the order of an eighth to a half inch per week, faster in dry heated air and where there is air movement across the grate. That range is the yardstick you compare a measured time-driven loss against. It is stated for a trap sitting still in occupied-building conditions, not for a drain in an unheated space or under a supply diffuser blowing across it.

Worked example: one gate, two units, opposite answers

Same building, same complaint, two units, both traps nominal 2.0 inches deep.

Unit 3B, vacant six weeks. Refilled to 2.0 inches.

• At 24 hours, no events: measured 2.0 inches. Change below the probe's 0.06 inch resolution, so it is reported as no measurable change rather than as zero.
• At 7 days, no events: measured 1.7 inches. Loss 0.3 inches over 7 days, so 0.3 inches per week.
• Against the yardstick: 0.3 sits inside the eighth to half inch per week evaporation range for a still trap in a conditioned space, and this unit is conditioned and still. Consistent with evaporation, no wick found when the trap was opened, so capillary action is out.
• Where it lands: from a full 2.0 inches at 0.3 inches per week, the seal takes about 6.7 weeks to reach empty.

Now check it against the story. The occupant reported the smell returning about three weeks after last running water. Three weeks at 0.3 inches per week is 0.9 inches lost, leaving 1.1 inches of seal. A short seal is still a seal, and it is still a barrier. So evaporation is real here, measured, and does NOT explain a smell at three weeks. Note also that the plus or minus 1 inch of water column system criterion is written against a nominal two-inch seal, so it stopped applying to this trap the moment it read 1.1 inches. That condition belongs to the trap-seal article and is not re-derived here; what matters is that a partly depleted seal has become vulnerable to an event it used to survive.

Unit 4A, occupied. Refilled to 2.0 inches.

• At 24 hours with a logged zero events: 2.0 inches, no measurable change. Time-driven loss is not the mechanism here.
• One logged event, the tub in the unit above dumping: re-dipped at 1.2 inches. Loss 0.8 inches per event.
• Convert to a comparable basis before saying anything about severity. At one such event per day, 0.8 inches per event times 7 events per week is 5.6 inches per week equivalent. Against unit 3B's 0.3 inches per week, that is close to twenty times the rate, on the same basis, from a single daily event.
• Direction, observed rather than inferred: the level was drawn down and air was heard pulling through the seal at the end of the upstairs discharge. Negative. Water was not lifted, nothing was thrown onto the floor, nothing bubbled in a bowl.
• Negative plus another fixture's discharge is induced siphonage. Not self-siphonage, because the fixture's own discharge produced no measurable change on a separate trial. Not back pressure, because the direction is wrong.

What each unit needs. 3B needs the seal maintained against clock time: a trap primer on a supply that actually runs, a deeper seal, or the honest answer that an unused drain in a unit nobody occupies will go dry and should be scheduled for refilling. A primer here is the right device. 4A needs an air path, because no amount of topping up survives 0.8 inches per event. A primer in 4A would refill a trap that is emptied again the next morning, and the customer would have paid for a device that makes the complaint intermittent instead of constant.

The failure mode. The error that produces the callback is running this test in one unit and applying the answer to the building. Both mechanisms were live in the same building at the same time, and they needed different work. A tech who dipped only 4A would have called the whole building a venting problem and left 3B's dry drain in place.

How to verify you separated the families rather than guessing

Re-run the gate with the event variable flipped, which is the only check that cannot be satisfied by a coincidence.

For a suspected event-driven loss, refill and then deliberately withhold the event for a full week. A trap that holds for seven days with zero discharges and then loses in a single logged event has been proven event-driven twice, once by the loss and once by the hold.

For a suspected time-driven loss, put the trap under load. Refill, then run the fixture and its neighbours normally for two days and re-dip. A time-driven trap comes back near full because each use refills it. That is the tell people miss: a trap losing to evaporation gets BETTER with occupancy, and a trap losing to siphonage gets worse, so the customer's own sentence about when it smells is diagnostic before you open anything.

Then close the loop on direction. If you called it positive, you should be able to find water outside the trap - a stained floor around a drain, a ring on the underside of a bowl rim, splash marks in the trap access. If you called it negative and there is standing water on the floor around the fixture, one of the two observations is wrong and it is usually the call, not the water.

References

• The plumbing code as adopted and amended by your local jurisdiction (the IPC or UPC family), which sets minimum trap seal depth and the venting requirements that limit system pressure, binding 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, before any entry into a manhole, wet well or interceptor while tracing a branch
• Safety data sheet for any drain-cleaning chemical the customer reports using, for the splash protection and glove material named before a joint is opened
• See related: Why a Trap Seal Is the Only Barrier and What It Is Made Of; What Siphonage Is and Why Only a Vent Stops It; What Back Pressure Does at the Base of a Stack