How to Work a Smell Complaint Back to a Mechanism

Why this matters

A sewer odor call has no part to swap and no reading that names the fault, so it gets worked from memory and closed on a guess: pour water in the floor drain, tell the customer it should be better, leave. Often it IS better for a week, which is worse than failing outright, because the shop now believes the guess worked. The way out is to stop treating the smell as the evidence and start treating it as one field in a short written record. This article is that record, built one field at a time, with the reason each field earns its place and what it removes from the list of things it could still be.

Before the record: the two smells that are not this call

If the odor is fuel gas, stop and leave. Natural gas and LP carry an added odorant that reads as sulfur or rotten egg to most people, which is exactly how sewer gas reads, and the two calls have opposite first moves. If there is any chance the smell is fuel gas: everyone leaves the building immediately, no light switches, no thermostat, no garage door opener, no phone used inside, and the call to the gas utility is made from outside and away. Do not stay to narrow it down. The library's gas-odor response article owns that branch.

If the trace would put you into a manhole, a lift station wet well or a large interceptor, that is a permit-required confined space and this record does not authorize entry. Entry requires a written 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, and a field-service shop can be under either Part depending on the job. Brief anyone with you, before you start, that nobody follows anybody in. Rescuer entry by an untrained bystander is how one casualty becomes several.

Everywhere below, keep your face out of open drains and cleanouts and work from the side. 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. It is the least reliable instrument on the job, and in any unventilated or partly enclosed space a calibrated multi-gas instrument is the only honest read.

Field 1: the odor class

Write down what it smells like, where, and what else in the building could produce it. This field exists to remove the entire non-drainage half of the problem in the first few minutes: a dead animal in a wall, a failed condensate pan, a dry-out product, a sump, a refrigerator drip tray, a floor finish curing.

Sewer odor has a wet, sour, sulfurous character and it comes and goes with building conditions. A dead animal is steadier and gets worse with heat. If the customer says it appeared the same week as new flooring, treat that as a live alternative rather than a coincidence.

Do not close this field on smell alone. Close it when you can also say which drainage opening is nearby.

Field 2: the timeline

Write the hours and the days, not "sometimes". Ask when it was first noticed, what time of day it is worst, whether it happens on weekends, and what the occupant was doing the last three times.

This field splits the mechanism list in half, and it is the highest-value few minutes on the job. A loss that tracks occupancy is event-driven. A loss that appears in an empty unit after a stretch of no use is time-driven. Those two need opposite repairs, and the seal-loss article covers the mechanisms and the controlled test that confirms which family you are in. Get the timeline before you touch a tool, because once you run water you have destroyed the state you were asked to explain.

Field 3: the location, with the height noted as a hint

Room, the fixtures in it, the branch it is on, and where in the room the smell is strongest.

Height is a weak cue and it must be written as one. Sewer gas is a mixture: hydrogen sulfide is heavier than air and tends low, methane is lighter and tends high, and an occupied room with any air movement in it mixes both. So "strongest at floor level" is a hint that points at a floor drain rather than a finding that proves one. It earns a line in the record and never carries a conclusion by itself.

Field 4: building operation at the time

Exhaust fans running, windows open, doors closed, wind, and whether the complaint follows any of them. A building held negative by its own exhaust pulls whatever is available through the weakest opening, and a drainage system is full of candidates.

This field is what turns "it only happens sometimes" into a testable statement. It is also the field most often left blank, and a blank here is why so many of these calls end at "intermittent, monitor."

Field 5: measured seal depths

Dip every trap on the branch, including the ones nobody uses, and write each depth with the resolution of your probe beside it. A marked probe or stiff tube on a rule reads to about a sixteenth of an inch, so 0.06 inches is the smallest change you are allowed to report. Anything smaller is noise and gets written as no measurable change, not as zero.

This is the first field with a number in it, and it is the field that converts the whole call from opinion to measurement.

Field 6: the event test and its direction

Refill each trap to full, then run one logged discharge and re-dip. Two numbers come out: the loss in inches per event, and the direction.

Direction is observed, never inferred. Level drawn down with air pulled through the seal is negative, and that is a gurgle. Water lifted and thrown toward the room, bubbling up through a bowl, wet floor around a drain, is positive, and that is a burp. They come from opposite ends of the system.

Field 7: the barrier walk

If every seal holds, the fault is not a seal. Walk the five real barriers in that space - piping and joints, trap seals, cleanout plugs, closet flange seals, and caps on abandoned openings - and photograph each one as you clear it. The barrier article enumerates them and explains why only one of the five can fail without leaving evidence.

Field 8: what is left

Write the residual list explicitly. A call closed with "found and corrected" and nothing in this field is a call that will come back, because nobody recorded what was ruled out.

Worked example: the record filled in on one call

Third-floor apartment, sewer odor.

Field 1. Wet and sulfurous, strongest in the hall bath. No recent finishes, no attic access above, nothing else in the room that drips. Sewer odor confirmed as the working class, non-drainage sources noted as unlikely rather than excluded.

Field 2. First noticed six weeks ago. Worst between about 6 and 9 in the evening on weekdays. Absent on weekends. That pattern tracks occupancy, so the working family is event-driven and the time-driven family drops down the list.

Field 3. Hall bath, strongest at floor level near the tub. Recorded as a hint pointing at the tub trap and the absence of a floor drain in this room, not as a finding.

Field 4. Building has a rooftop exhaust that runs on a schedule and shuts down overnight. Occupant keeps the bath door closed. Nothing yet ties the complaint to fan operation.

Field 5. Nominal seal depth on all three traps, 2.0 inches. Probe resolution, 0.06 inches.

• Lavatory: 2.0 inches
• Water closet: full
• Tub: 0.4 inches

One trap out of three is short. That is the finding the rest of the call is built on.

Field 6, the event test. Tub trap refilled to 2.0 inches.

• Tub's own discharge, run alone: re-dipped at 1.95 inches. The 0.05 inch change is under the 0.06 inch resolution, so it is recorded as no measurable change, and self-siphonage is not demonstrated.
• One logged discharge from the unit above, nothing else running: re-dipped at 1.1 inches. Loss 0.9 inches per event.
• Direction, observed at the fixture: level drawn down, air heard pulling through at the end of the upstairs discharge, nothing lifted, no water on the floor. Negative.

The basis conversion, printed rather than assumed. Inches per event is not comparable to a complaint window until it meets an event rate. Three upstairs discharges across an evening at 0.9 inches each is 2.7 inches of equivalent loss against a 2.0 inch seal, so the seal is gone before the evening is over and stays gone until somebody runs the tub. That is the arithmetic behind the 6 to 9 p.m. window and behind the weekend absence, when the building above is lightly used.

One condition carried forward rather than re-derived. The plus or minus 1 inch of water column system criterion is written against a nominal two-inch seal, so it stopped describing this trap once it read 1.1 inches. A depleted seal fails events it used to survive, which is why the complaint got worse over six weeks rather than appearing all at once. The trap-seal article owns that condition.

Field 7. Not required. A seal was demonstrably being removed, so the barrier walk was recorded as not performed rather than as clear.

Field 8, the residual. Negative event-driven loss at the tub, induced by discharge from above, self-siphonage not demonstrated. Two live candidates remain: an inadequate or blocked air path serving this branch, or a branch running deep enough at peak that the flow seals the vent connection along it. Those are the siphonage article's territory and the next visit's scope.

The failure mode this record prevents. Without field 6, the tub reading of 0.4 inches looks like a dry trap, and the standard wrong move is to fill it, tell the customer to run water weekly, and leave. That fix is correct for a time-driven loss and useless here: at 0.9 inches per event the trap is empty again the first evening the upstairs unit showers. The customer then reports that the repair lasted one day, and the shop has spent a visit proving nothing.

How to verify the record is finished rather than abandoned

Read it back and check three things.

Every field has either a value or an explicit "not performed". A blank is indistinguishable from a zero to the next tech, and fields 4 and 7 are the two that go blank.

The timeline and the arithmetic agree. If field 2 says weekday evenings and field 6 produces a loss rate that would empty the trap in a week regardless of occupancy, one of them is wrong and you have not finished. In the example they agree, which is what makes the conclusion worth acting on.

The residual names a mechanism, not a location. "Smell in the hall bath" is where you started. "Negative event-driven loss at the tub trap, induced from above" is a mechanism, and it tells the next tech what to bring: a camera or a smoke machine and a look at the branch venting, rather than another jug of water.

If the next step is a smoke test, notify every occupant in the affected area in advance and in writing, read the smoke fluid's safety data sheet for the ventilation it calls for, and keep it out of any unit whose occupant has told you about a respiratory condition. If the next step is a look at the vent terminal, fall protection on the roof is required and it forks by Part: 29 CFR 1910.28 in general industry, 29 CFR 1926.501 in construction.

References

• 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 encountered while tracing a branch
• 29 CFR 1910.28 (general industry) and 29 CFR 1926.501 (construction), fall protection for roof access to a vent terminal
• Safety data sheet for the smoke fluid used in any smoke test, for its ventilation and occupant-notification requirements
• The plumbing code as adopted and amended by your local jurisdiction (the IPC or UPC family), which sets trap seal depth and the venting that limits system pressure, binding through the adopting ordinance and the permit rather than on its own
• See related: How a Trap Seal Is Lost and How to Tell Which Way It Went; Why a Trap Seal Is the Only Barrier and What It Is Made Of; What Siphonage Is and Why Only a Vent Stops It