How to Run a Smoke Test Without Creating a Complaint

Why this matters

A smoke test pushes tracer smoke through a sewer at a pressure far below what a trap seal can hold, which is exactly why it can be run on an occupied street at all. The arithmetic behind that sits in the sibling card on pressure testing, and its consequence is the whole basis of this procedure: smoke inside a building is not something the test did, it is something the test revealed, provided the blower stayed under a trap seal's rating. Held there, every indoor complaint is a barrier that was already missing, and the occupant has been breathing through that gap since before your truck arrived. Raise the pressure to chase a stronger signal and you own the smoke yourself, which is why the pressure you actually held is the last thing this card asks you to record. So the work that decides whether the day produces findings or complaints happens before the blower starts, and the steps below are ordered by the lead time each one needs, longest first, because the ones that need a week cannot be recovered on the morning of the test.

The controls that come before any of it

The blower is an inhalation hazard in two ways. If it is engine driven it makes carbon monoxide, so the exhaust gets positioned downwind and away from the manhole opening and from every building opening, and the operator wears a personal CO monitor rather than relying on being outdoors. The smoke fluid itself is an irritant: read its safety data sheet before the first fill, under the hazard communication standard at 29 CFR 1910.1200, work upwind of the plume, and anyone who has to enter a smoke-filled space does so under a respiratory protection program meeting 29 CFR 1910.134. A dust mask is not respiratory protection and does nothing for a vapour.

Use a liquid-smoke blower, not a pyrotechnic smoke candle, anywhere the smoke enters a sewer. A sewer headspace can hold methane. Introducing a burning device into it is an ignition source in a potentially flammable atmosphere, and the fact that it has usually been fine is not a control.

The setup is at a manhole, which is a permit-required confined space whether or not anyone intends to enter: 29 CFR 1910 Part 1910.146 in general industry, 29 CFR 1926 Subpart AA on construction work. Name which Part your job falls under. Hydrogen sulfide deadens the sense of smell well below the concentrations that hurt you, so nobody uses their nose as the instrument at an open lid. Lids come off with a proper hook, feet clear of the swing.

Work in a roadway carries a traffic control plan conforming to the MUTCD as adopted by your state, with high-visibility apparel on everyone in the work area; on construction work the signs, signals and barricades requirements sit in 29 CFR 1926 Subpart G.

Seven to fourteen days out: the notifications that cannot be rushed

Notify the fire department and the emergency dispatch centre in writing, with the date, the window, the streets and a crew contact number. Smoke coming out of a foundation looks exactly like a structure fire to a neighbour, and a fire response to your test is the worst possible outcome of the day: it puts a crew at risk, it burns a public resource, and it makes the local news.

In the same notice, open a vulnerable-occupant line. You are looking for home oxygen users, people with significant respiratory conditions, bedridden residents and anyone who cannot leave a building quickly. That list takes days to come back because it depends on people reading a letter and calling.

Skipped: the fire department finds out from the first 911 call, and the one address that needed a different plan is the one address you will hear about afterwards.

Three to five days out: the written notice to every address

Door hanger or letter to every address in the test area, in the languages actually spoken there. It says what will happen, when, that the smoke is a tracer rather than a fire, what to do if smoke appears indoors, and the crew's number. It also carries one instruction that does real work: run water in every drain that has not been used lately, naming the ones people forget, floor drains, basement laundry standpipes, spare bathrooms, utility sinks.

That instruction is doing your pre-fill for you in the units you will never get access to. It is the highest-yield sentence in the notice and it belongs in bold.

Skipped: every dry trap in every occupied unit becomes an indoor smoke event, and you will spend the afternoon proving to people that you did not fill their apartment with smoke.

One to two days out: confirm rather than assume

Confirm the fire department has the notice on the run card for those streets, not just in an inbox. Confirm the vulnerable-occupant list and decide what each entry needs: a schedule change, a pre-fill visit, or an agreement that they are away during the window. Stage the crew number where the answering phone is actually staffed on the day.

Skipped: the notification exists and does nothing, which is the worst of both worlds because it also makes you feel prepared.

The morning of: inventory the unused traps and the open holes

Walk the test area and build two lists. First, every trap you can reach that is not in regular use: vacant units, common-area floor drains, mechanical room drains, basement drains. Second, every opening that is already unsealed: a cleanout with no cap, an abandoned stub, a broken frame seal on a structure. The second list is what the test is going to find anyway; recording it now separates known-before from found-during in the report.

Where you can reach a dry trap, fill it. Where you cannot, record the address and the reason.

Skipped: you cannot tell afterwards whether an emergence point was a defect you found or a hole you already knew about, and the report loses the distinction that makes it useful.

Minutes before: observers, wind, and the pressure you will hold

Observers in position with the streets divided so nobody is watching two blocks at once, and told to watch buildings and not only the ground. Note the wind, because a light plume drifting off a foundation is easy to miss downwind and easy to misattribute upwind. Set the blower to hold a low, steady pressure on the order of a quarter inch of water column, which is roughly an eighth of what a two-inch trap seal can hold. The sibling card on pressure testing carries that arithmetic; do not raise the pressure to get a better result, because above the seal's rating you are no longer running a smoke test on a live system, you are blowing seals.

Skipped: a crew chasing a stronger signal turns a diagnostic into the exact event the notifications were meant to prevent.

During and immediately after

During: one person on the phone, everyone else observing, every emergence point photographed with a fixed reference and located. Positions are the entire output of this test, so a photograph with no landmark in it is a wasted finding.

After: shut down, ventilate anything that took smoke inside before anyone re-enters, and go to every address that reported indoor smoke before leaving the area. That visit is not customer service, it is the diagnosis: something in that building has no barrier, and you are standing there with the exact information needed to find it.

Worked example: 60 addresses in one morning

Test area: four garden-apartment buildings totalling 42 units, plus 18 single-family homes. Sixty addresses.

Notification. Fire department and dispatch notified 10 days out. Door hangers to all 60 addresses 4 days out. Three responses to the vulnerable-occupant line: one home oxygen user, two residents with asthma. The oxygen user's building was scheduled for the end of the window with a crew member assigned to that address specifically.

Unused-trap inventory on the morning of. Twenty-eight traps identified as not in regular use:

• 6 vacant units at 3 traps each: 18
• Common areas, 2 laundry floor drains and 3 mechanical room drains: 5
• Basement floor drains, one per building plus one in a storage room: 5

Total identified: 18 + 5 + 5 = 28. Filled: 27. Not filled: 1, a basement floor drain in a locked storage room where no key could be found before the window opened. Two crew members, about 1.5 hours, so 3 labour hours of pre-fill.

Known-unsealed inventory on the morning of. Four openings photographed before the blower ran: two cracked cleanout caps, one abandoned stub with no cap, one broken frame seal on a yard structure.

Result. Eleven emergence points outdoors. Four were on the morning's known-unsealed list and are not discoveries. Seven were found by the test, including a downspout tied into the sanitary line, which is the discrete inflow connection the whole exercise was for.

One indoor report, and it came from the building with the locked storage room. That is the honest finding and it is worth stating plainly rather than rounding off: 27 of 28 identified traps were filled, and the indoor smoke appeared at the one that was not. The mechanism did exactly what the sibling card's arithmetic says it does. That floor drain had been dry for long enough that the smoke found an open path, which means sewer gas had the same open path every day before the test.

What the pre-fill was worth, in comparable hours. On the previous block, run without a pre-fill pass, four indoor reports each cost a return visit and a written response, about 5 crew hours in total. Here the pre-fill cost 3 crew hours and one report cost about 1.2. Both figures are unbilled crew effort, so they compare directly: 3.0 plus 1.2 against 5.0, a modest saving in hours. The real difference is not the hours. It is that four occupants on the previous block learned about a missing barrier by having their apartment fill with smoke, and here one did.

Failure mode. The version that goes wrong is a crew that treats the notification as the whole preparation. The letters go out, the fire department is told, the blower runs, and nobody walks the vacant units. The test finds its inflow connections and also produces nine indoor reports, the client fields the calls, and the next test in that jurisdiction gets scheduled with a level of public suspicion that no amount of door hangers fixes.

Closing the day out

• Reconcile your emergence points against the morning's known-unsealed list and report the two categories separately. A cleanout cap you photographed at 8 am is not a discovery at 11 am.
• Count the indoor reports and check the number against your own inventory of unfilled traps. If there are more indoor reports than unfilled traps, you have found barriers missing for reasons other than dryness, and each of those is its own defect worth a line.
• Verify every emergence photograph has a locatable reference in frame. A position you cannot return to is not a position.
• Record the addresses you could not access and why. That list is the schedule for the next visit, and it is the honest answer when somebody asks why one building reported smoke.
• Confirm the blower pressure you actually held, not the one you intended to hold. A test run above a trap seal's rating produced its indoor smoke itself, and the report has to say so.

References

• 29 CFR 1910 Part 1910.146, permit-required confined spaces, general industry; 29 CFR 1926 Subpart AA, the construction counterpart
• 29 CFR 1910.1200, hazard communication, for the smoke fluid safety data sheet; 29 CFR 1910.134, respiratory protection, for anyone entering a smoke-filled space
• 29 CFR 1926 Subpart G, signs, signals and barricades, on construction work, with traffic control conforming to the MUTCD as adopted by your state
• See related: Why a Pressure Test and a Smoke Test Find Different Faults; What Inflow and Infiltration Look Like From the Building Side; Why a Trap Seal Is the Only Barrier and What It Is Made Of