Measuring flue draft: where, when and what it means

When did you last take a draft reading with the mechanical room door shut, the laundry dryers running and the ground-floor restaurant hood at full speed? Almost nobody does. Yet that is the configuration in which the appliance spends its worst hours of the year. Measuring flue draft takes two minutes; measuring it under the conditions the building actually creates takes ten more — and those ten minutes decide whether combustion gases go up the chimney or fall back into the room.

Draft is the smallest quantity a technician records all day. On a natural-draft appliance you are looking for a few hundredths of an inch of water column, negative — the figure commonly cited for Category 1 appliances is around -0.02 in. w.c. Since one inch of water column equals 249 Pa, that is barely 5 Pa. Which means the instrument, the test point and the timing of the reading matter more here than anywhere else in the boiler room.

Tools you need

• Low-range differential manometer (±1 to ±2 in. w.c.), 0.01 in. w.c. resolution, or a direct readout in pascals.
• Rigid draft probe with flexible tubing, long enough to clear the flue wall and any insulation.
• Drill and 1/4-inch bit, plus a metal plug or self-tapping screw to close the hole.
• Calibrated combustion analyzer: flue CO belongs in the same visit.
• Personal CO monitor worn on you, separate from the analyzer.
• The outdoor temperature at that hour — weather reading or thermometer. Without it, the draft number has no context.

⚠ Safety: what stops the test immediately

Visible spillage at the draft hood is not a measurement to document, it is a condition to correct: shut the appliance down, ventilate, find the cause. Same if your personal monitor rises in ambient air. Quebec’s occupational exposure limits for carbon monoxide — 35 ppm as an eight-hour weighted average and 175 ppm over fifteen minutes under the RSST — describe a worker exposure ceiling, not a tolerance for a mechanical room: any measurable ambient CO in a boiler room signals a venting problem to resolve before going further.

1. Where to drill the test hole — and why only one

The probe has to see what the appliance sees, not what the mixture became after dilution. The test point sits downstream of the draft hood or barometric damper, one to two feet along the connector, or at least two flue diameters from the appliance outlet. Closer, and turbulence from the elbow corrupts everything; farther, and dilution air has already entered, giving you a flattering number that describes nothing.

One hole, drilled once, closed with a metal plug. Flue connectors in older Montreal boiler rooms that have seen fifteen technicians over twenty years sometimes look like colanders, and every forgotten hole is a parasitic air inlet degrading the very draft the next technician will try to measure.

2. Zero the gauge in the boiler room, not in the truck

At a scale of a few pascals, thermal drift in a pressure sensor stops being a theoretical concern: it is the same order of magnitude as the value you are after. Between a toolbox in a service van on a January morning and a heated mechanical room lies some forty degrees of difference — and the sensor carries that gap into its first reading with nothing to announce it. So take the gauge out on arrival, walk the equipment while it acclimatizes, then re-zero it in the room with both ports open. Choosing the instrument itself is a separate discipline, covered in our digital manometer guide: a ±60 in. w.c. gauge simply does not have the resolution this job requires.

3. Measuring flue draft at steady state: the reference reading

The first number is worthless. At start-up the chimney is cold, its column of air still weighs as much as the outdoor air, and draft builds only as the flue gases warm it. ANSI standards allow the appliance five minutes to prove draft; inspection protocols treat spillage beyond 60 seconds after ignition as a failure.

So record two pieces of information, not one: the steady-state value, and how long it took to get there. An appliance that needs four minutes is still within the standard, but it is telling you it works at the edge of its envelope. The day outdoor temperature, occupancy or an exhaust fan changes, that is the appliance that tips first.

4. The test almost nobody runs: worst-case depressurization

This is where a measurement becomes a diagnosis. Natural draft is a contest between two forces: the chimney’s buoyancy and the building’s negative pressure. Testing the first without loading the second is like testing brakes on a flat parking lot.

The protocol is three moves: close the mechanical room doors and the doors separating it from the rest of the building, switch on everything that removes air — commercial hoods, dryers, bathroom fans, garage exhaust, ventilation systems at full flow — then repeat the draft and ambient CO readings. If draft collapses, you have just reproduced the back-drafting condition deliberately and under control. Then open an exterior door: if everything returns to normal, the cause is depressurization, not the chimney.

That test points straight back at combustion air requirements. The RBQ’s technical directive on combustion air supply requires the installer to notify the owner in writing when the openings are inadequate for all appliances sharing an enclosure, and its companion directive on ventilation air openings sets the trigger at any gas appliance above 400,000 BTU/h, or 120 kW. Sizing falls under Chapter II, Gas, of the Construction Code and CSA B149.1 — the subject of our article on combustion air requirements.

5. Why does draft fail only in deep cold?

The usual instinct says the opposite: the colder it gets, the better the draft should be, since the temperature difference between flue gases and outdoor air increases buoyancy. That is true of the chimney. It is equally true of the building.

A multi-storey building is itself a chimney. In January it draws cold air in low — doorways, service shafts, elevator cores — and pushes it out high. The basement, where most of Montreal’s older building stock keeps its boiler room, sits at the most negative point of that gradient. The minimum draft thresholds used in inspection protocols reflect exactly this: they tighten as outdoor temperature drops, calling for roughly -2.5 Pa below -12 °C, against a fraction of a pascal in mild weather.

The practical consequence, and the heart of the matter: a draft test passed in September says nothing about January. On a sensitive building, the reference reading gets repeated once in cold weather, on a windy day, with ventilation running as it normally does.

Field case: a Laval building and a commercial dryer

Six-storey residential building in Laval, two atmospheric gas boilers sharing a masonry chimney, basement mechanical room. Recurring winter complaint: a smell of combustion gases on the ground floor, two or three times a winter, never reproducible during a visit.

Steady-state draft, doors open, a November afternoon at 4 °C: within spec, established in under a minute. Case closed, apparently. The worst-case depressurization test took six minutes: doors shut, both commercial dryers in the basement laundry running, the waste-room exhaust fan on — draft went below zero and the mirror at the draft hood fogged. Ambient CO moved.

The chimney was not the cause. The laundry had been refitted two years earlier; two commercial dryers had replaced three domestic machines, with no make-up air, in a room communicating with the boiler room. The fix was an air inlet sized for all the appliances in the basement, with draft readings before and after, doors closed and exhaust fans running. The following winter brought no complaints.

What to put in the report next time

On the next commissioning or service visit to a natural-draft appliance, ask for four lines in the report instead of one: steady-state draft, worst-case draft, time to establish, and the outdoor temperature at the moment of the reading. Those four lines fit in a single table row, and they compare year over year — which “draft OK” never will.

On the management side, the gap between those two formats is counted in emergency calls avoided: one line saying “within spec” covers the visit; four lines describing how the appliance behaves under load let you anticipate a back-draft before the complaint arrives, and discuss make-up air with numbers instead of instinct. That is the basis on which Montréal Combustion recommends — or rules out — venting work inside a preventive maintenance program, and the kind of record you want on file when an appliance changes hands.