How Draft Actually Works

Why this matters

Draft gets talked about as if it were a property of the chimney, something a masonry stack either has or lacks. It is not. Draft is a pressure difference produced right now by two temperatures, a height and whatever the building is doing, and it can disappear from a vent that worked for twenty years with nothing about that vent having changed. The most common version of that in the field is an appliance that starts spilling the winter after an unrelated piece of equipment was replaced, and the tech who inspects the chimney and finds it sound, clear and correctly built has verified the one thing that was never the problem.

Before you put anything into a flue

Have a carbon monoxide monitor running in the space before you fire the appliance. Carbon monoxide is odourless and colourless at every concentration that will hurt you. If it climbs toward the OSHA general industry permissible exposure limit of 50 ppm as an 8-hour time-weighted average at 29 CFR 1910.1000 Table Z-1, or anyone reports headache, nausea or confusion, everyone leaves the building immediately, nobody operates a switch on the way out, ventilate from outside, and nobody re-enters until an instrument reads it clear. If you smell gas, that is the other instruction: everyone out immediately, no switches, no lights, no phone used inside, call from outside.

If you need a draft test port, drill only metal vent and wear eye protection against chips. Never drill a plastic vent on a condensing appliance, which is a listed sealed pressurised system with its own designated port. Let a probe cool before you handle it, or use gloves rated for contact heat, because a probe pulled from a flue will cause a deep contact burn.

And if confirming the vent termination means getting on the roof, treat that as its own hazard rather than a five-minute look. Fall protection duties live in both places a field-service shop can fall under: 29 CFR 1910.28 for general industry, which triggers at 4 feet above a lower level for most walking-working surfaces, and 29 CFR 1926.501 for construction, which triggers at 6 feet. Use the required protection for whichever Part your work falls under, or get eyes on the termination from the ground or from a drone rather than free-climbing a pitched roof to answer a question you can often answer another way.

What actually produces draft

Hot flue gas is less dense than the outdoor air at the same elevation. Stack a column of it inside a vent and the weight of that column is less than the weight of an equal-height column of outside air. The difference shows up as a lower pressure at the base of the vent than at the appliance's air inlet, and that difference is what pulls products up and out.

Two variables set it in practice, with a third that matters if you work at elevation:

• Height. More vertical run means more column, so more difference. This is why a short vent on a single-storey house is more fragile than a tall one on a three-storey.

• The temperature difference between the flue gas and the outdoor air, expressed on an absolute scale. Draft is proportional to height times the quantity (1 divided by outdoor absolute temperature) minus (1 divided by flue absolute temperature).

• Barometric pressure, which means altitude. Theoretical draft scales with air density, so it falls with elevation: at 6,000 ft you have roughly 80 percent of the sea-level pressure and roughly 80 percent of the draft from the same stack at the same temperatures. At sea level you can ignore this. Above about 2,000 ft you cannot, and a vent sized on a sea-level rule of thumb is one of the quieter ways a mountain-town appliance ends up marginal.

Everything else that affects draft, and there is plenty of it, works by changing one of those three or by adding resistance the difference has to overcome.

The two temperatures, and which one you can actually move

Take a 20 foot vent and run the arithmetic in degrees Rankine, which is degrees Fahrenheit plus 460.

Case one, a cold day. Outdoor 40 degrees F, or 500 R. Flue gas 300 degrees F, or 760 R. The temperature term is 1 divided by 500 minus 1 divided by 760, which is 0.002000 minus 0.001316, or 0.000684.

Case two, a mild day, same appliance. Outdoor 60 degrees F, or 520 R. Flue still 300 degrees F. The term is 1 divided by 520 minus 1 divided by 760, which is 0.001923 minus 0.001316, or 0.000607.

That 20 degree F swing in outdoor temperature changes theoretical draft by 0.000684 divided by 0.000607, or about 13%. Real, measurable, and the reason a marginal vent misbehaves in shoulder season and behaves in January.

Case three, the one that matters. Cold day again, outdoor 500 R, but the flue gas has cooled to 160 degrees F, or 620 R, by the time it reaches the top. The term is 0.002000 minus 0.001613, which is 0.000387. Against case one that is a 43% loss of theoretical draft.

Cooling the flue gas by 140 degrees F cost more than three times what a 20 degree F outdoor swing cost. Flue gas temperature is the lever, and it is the one that field changes actually move.

Draft is a property of the operating system, not of the vent

Four things routinely destroy draft with no change to the vent itself:

• Losing heat input into the vent. A vent sized for two appliances that now serves one, or an appliance replaced with a more efficient one, puts less heat into the same flue area. The gas moves slower, the vent walls stay cold, the gas cools further, and draft falls in a loop that feeds itself.
• An oversized or cold vent. Same mechanism from a different starting point. An exterior masonry chimney on a north wall takes far longer to warm than an interior one, and until it warms it is a heat sink rather than a chimney.
• The building pulling harder than the vent. Exhaust equipment can put a mechanical room negative enough that the path of least resistance for make-up air runs down the vent. That half of the problem, including how to test it and what limits to work to, belongs to the combustion air article and is not re-derived here.
• Wind and termination effects. A termination in a pressure zone created by a nearby roof, wall or adjacent structure can produce positive pressure at the outlet under certain wind directions, which is why a spillage complaint that correlates with weather is a real pattern and not a customer's imagination.

The draft survey, filled in

An orphaned atmospheric water heater, 40,000 Btu/h, on a masonry chimney that formerly common-vented it with a 100,000 Btu/h furnace. The furnace was replaced eighteen months ago with a condensing unit vented separately in plastic. Nothing was done to the chimney.

Draft measured at the breech, in inches of water column, from a genuine cold start. Negative means the vent is drawing; positive means it is not.

Atmospheric natural-draft appliances commonly want somewhere in the range of -0.02 to -0.04 inches of water column at the breech, but that band is a starting point, not a specification: read the value the appliance manufacturer states and work to that.

As found, the vent was positive for the first full minute, meaning products were going into the room rather than up. It then went weakly negative, and the strongest it ever achieved was -0.005 inches of water column, which is a quarter of the low end of that band. It never recovered further across ten minutes.

What the shape of that survey said

Read the two things the numbers do that a single reading cannot.

It started positive. Some spillage in the first seconds of a cold start is normal while a vent warms. A full minute of it is not, and it means the vent is starting from so far behind that the appliance's own heat cannot dig it out.

It strengthened, and it strengthened nowhere. From three minutes to ten minutes it gained 0.002 inches of water column, going from -0.003 to -0.005, which is real movement in the right direction and still leaves it at a quarter of the low end of the band. A vent that is climbing steeply at ten minutes has a slow warm-up problem, which has a different fix. This one was climbing at a rate that would not reach the band inside any run this appliance ever makes.

The cause follows straight from the input arithmetic. The chimney was sized for 140,000 Btu/h of combined input and now receives 40,000, which is 29% of the heat it was designed around going into the same flue area. Roughly 22 volumes of hot products leave an appliance for every volume of fuel, and the volume matters as much as the temperature; that relationship is worked out in the combustion fundamentals article. With less than a third of the heat, the products slow, they give up their temperature to a large cold masonry mass, and the third case in the arithmetic above is exactly what happens: the flue gas arrives at the top too cool to lift itself.

The correction, and what it was not

The fix was a liner sized for the 40,000 Btu/h appliance alone, run continuously to the termination, per the vent sizing tables in the National Fuel Gas Code, NFPA 54, using the table for the appliance category and vent configuration actually present.

After relining, the survey went negative on the first reading and did most of its work early: it reached -0.028 inches of water column by three minutes and gained only 0.004 more across the following seven, settling at -0.032, inside the manufacturer's stated range.

What the fix was not: a draft inducer bolted on to force the issue. A powered inducer is a legitimate solution in some installations, but adding one to an oversized cold flue treats a symptom and leaves a vent full of condensing flue gas that will damage the masonry. It also introduces a device that must be interlocked so the appliance cannot fire without proven draft, and an inducer installed without that interlock is worse than the original condition.

What it also was not: a bigger appliance. Restoring heat input to a chimney by oversizing the water heater is a real temptation and it solves the draft complaint by creating a short-cycling, oversized appliance.

What changes the answer

Fan-assisted and sealed-combustion appliances. These do not rely on buoyancy the same way, and the numbers above do not transfer. Their vent systems are pressurised or mechanically driven, are sized from entirely different tables, and are tested against manufacturer-stated pressure values rather than a natural-draft band.

Oil-fired equipment. Over-fire draft is measured in addition to breech draft, and the relationship between the two is the diagnostic. Do not substitute one for the other.

Barometric dampers. A vent fitted with a barometric damper will read a regulated draft at the breech that partly hides what the chimney is doing, because the damper is admitting room air to hold the setpoint. Measure both sides of it and understand that the damper's job is to limit excess draft, not to create draft that was never there.

How to verify you got this right

Verify a draft correction on a cold start, in the coldest weather you can get, with the building in its worst-case exhaust condition, and take the reading as a series rather than a single number. Draft measured on a mild afternoon with the mechanical room door open, ten minutes into a run, is the easiest test the installation will ever face, and it is the one most likely to be recorded.

Then confirm the outcome rather than the input: check for spillage at the draft hood after the burner has run 5 minutes in that same worst case, and confirm that any spillage present at ignition has cleared. A vent that reads inside the band and still spills has something else going on, most often on the building pressure side, and a vent that reads slightly outside the band and never spills under worst case is a different conversation from one that spills.

Write the whole series into the job record with the outdoor temperature and the building condition. A single draft number with no conditions attached cannot be compared to anything later, which means the next tech starts over.

References

• NFPA 54 (ANSI Z223.1), National Fuel Gas Code, venting provisions and vent sizing tables; use the edition adopted locally
• Appliance manufacturer documentation for the required draft value and measurement location
• 29 CFR 1910.28 (general industry, 4 foot trigger) and 29 CFR 1926.501 (construction, 6 foot trigger), fall protection duties
• 29 CFR 1910.1000 Table Z-1, general industry permissible exposure limit for carbon monoxide
• See related: Why Combustion Air Is a Quantity and Not an Opening; How Combustion Actually Works