Why Combustion Air Is a Quantity and Not an Opening

Why this matters

A grille is a hole. Air is a flow. The code sizes the hole so the flow can happen, but the hole does not produce the flow, the pressure difference across it does. Once a building has enough exhaust in it to overwhelm that pressure difference, a perfectly compliant opening delivers nothing and a natural-draft appliance starts pulling its combustion air backwards down its own vent, along with everything already in the flue. Techs who measure the grille and stop have verified a necessary condition and called it a sufficient one, and the difference between those two things is a carbon monoxide event.

When an appliance is spilling, this comes before the tape measure

If you confirm flue gases spilling into an occupied space, that is not a finding to write up and schedule. Shut the appliance off at its own gas shutoff, get the occupants out of the space, ventilate it from outside by opening exterior doors and windows, and leave the appliance out of service until the spillage is corrected. A red tag on a spilling appliance is the right outcome and it is not an overreaction.

If ambient carbon monoxide in the occupied space is climbing, or anyone reports headache, nausea or confusion, everyone leaves the building immediately, nobody operates a switch on the way out, and nobody re-enters until an instrument reads it clear. OSHA's general industry permissible exposure limit is 50 ppm as an 8-hour time-weighted average at 29 CFR 1910.1000 Table Z-1, with a NIOSH ceiling of 200 ppm. Have a carbon monoxide monitor running in the space before you fire anything, because the gas is odourless at every concentration that hurts you and you will not know otherwise.

If you smell gas rather than suspecting flue spillage, that is the other instruction: everyone out immediately, no switches, no lights, no phone used inside, shut off at the exterior shutoff only if it is on your way out, call from outside.

The call

A house with a 100,000 Btu/h furnace and a 40,000 Btu/h water heater in a small mechanical room, combined input 140,000 Btu/h. The homeowner reports a faint exhaust smell in the hallway on winter evenings and a water heater that "sometimes smells hot." A remodel eighteen months earlier added a high-capacity kitchen range hood and moved the clothes dryer from the garage into a laundry closet off the same hallway. Nothing was touched in the mechanical room and nothing about the appliances changed.

That last sentence is the whole case. When a combustion appliance starts misbehaving and nobody touched it, the change is in the building.

What the tape measure said

The mechanical room measures 10 feet by 12 feet by 8 feet, which is 960 cubic feet.

The National Fuel Gas Code, NFPA 54 (also published as ANSI Z223.1), sets the indoor-air standard method at a minimum of 50 cubic feet of space per 1,000 Btu/h of combined appliance input. For 140,000 Btu/h that is 140 times 50, or 7,000 cubic feet required. The room has 960. That is 14% of what the standard method requires, so this is a confined space and it needs openings to a source of air outside itself. Nobody is surprised by that, and openings were in fact installed.

The two-permanent-opening method to outdoors requires each of two openings to have a minimum of 1 square inch of free area per 4,000 Btu/h of total input, with one opening within 12 inches of the top of the enclosure and one within 12 inches of the bottom. For 140,000 Btu/h that is 140,000 divided by 4,000, or 35 square inches of free area per opening.

Two 6 inch by 6 inch grilles were installed, one high and one low, which is 36 square inches each. Gross. A tech with a tape measure reads 36 against a required 35, calls it compliant, and goes looking somewhere else.

The free-area trap

The code word is free area, not opening size. A louvered grille blocks a large fraction of its own gross area with the louvers themselves. Where the free area is not marked on the grille, NFPA 54 gives ranges rather than single figures, roughly 60 to 75% for a metal louver and 20 to 25% for a wood one. Use the LOW end when the free area is unmarked. On a combustion-air calculation the assumption bias has to run toward more opening rather than less: at 60% the grilles in the example below deliver about 21.6 in2 against the 35 required, not the 27 that the top of the range would suggest, and the difference decides whether the appliance gets its air.

Run it: 36 square inches gross times 75% is 27 square inches of free area. Required is 35. The installation is 8 square inches short per opening, which is 8 divided by 35, or 23% under. Had those grilles been wood, at 25% free area they would have delivered 9 square inches against a required 35, roughly a quarter of requirement.

To get 35 square inches of free area from a metal louver you need 35 divided by 0.75, or 46.7 square inches gross. An 8 inch by 6 inch grille is 48 square inches gross and 36 square inches free, which clears it. Both grilles were replaced.

And the appliance still spilled.

The half the tape measure cannot reach

Air does not move because there is an opening. It moves because there is a pressure difference across the opening. The sizing arithmetic above assumes the appliance's own draft, plus the small pressure differences a house normally carries, are enough to pull air in through that free area. Put enough exhaust into the building and that assumption stops being true.

Measure it. With everything off and the exterior doors closed, the mechanical room sat about 0.5 Pa negative with respect to outdoors, which is unremarkable. Then run the worst case: range hood on high, clothes dryer running, both bathroom fans on, and the interior door to the mechanical room closed, which is how the homeowner actually leaves it. The room went to about 8 Pa negative.

Building-performance practice commonly uses a worst-case depressurisation limit on the order of 3 Pa for atmospheric natural-draft appliances, with larger allowances for fan-assisted equipment; confirm the specific limit in the protocol your jurisdiction or program uses, because the number is protocol-specific rather than universal. Whatever limit you use, 8 Pa negative is not close to it.

Confirm it at the appliance rather than trusting the manometer alone. NFPA 54's spillage check looks for products of combustion escaping at the draft hood relief opening after the main burner has run for 5 minutes. In the as-found worst case, spillage at the water heater's draft hood was continuous well past 5 minutes. With the mechanical room door open it cleared in under a minute. That contrast is the diagnosis: the appliance is fine and the building is the fault.

The number the openings were never sized against

Here is the comparison that makes this intuitive.

Combined input is 140,000 Btu/h. At roughly 1,000 Btu per cubic foot of natural gas that is 140 cubic feet of gas per hour. Each cubic foot of gas needs about 9.57 cubic feet of air at exact stoichiometry, and appliances are run with a surplus above that; the stoichiometry behind those figures is worked through in the combustion fundamentals article. At about 45% excess air the two appliances together need roughly 1,940 cubic feet of air per hour, which is about 32 CFM.

Now total the exhaust: a high-capacity range hood on the order of 300 CFM, a dryer on the order of 150 CFM, two bath fans at roughly 50 CFM each. That is about 550 CFM of air leaving the building.

The exhaust devices move roughly 17 times the air the combustion appliances need. The grilles were sized for 32 CFM under gentle pressure. They were never a make-up air path for 550 CFM, and no compliant grille size would have made them one. The house found its make-up air where the resistance was lowest, which was down the water heater's vent.

What changes the answer

Sealed-combustion and direct-vent appliances. These draw combustion air through their own dedicated pipe from outdoors and are not affected by room pressure in the same way, which is why replacing an atmospheric water heater with a sealed unit is a legitimate fix for this exact fault rather than a dodge. It is often cheaper in labour hours than engineering make-up air for the whole house.

Fan-assisted appliances. An induced-draft appliance tolerates more depressurisation than an atmospheric one, so a house that spills a natural-draft water heater may run its fan-assisted furnace without complaint. Do not read the furnace's good behaviour as evidence the room is fine.

Two-rooms-communicating and mechanical combustion air methods. Where outdoor openings are impractical, the code provides alternatives, including permanent openings between adjacent rooms and powered combustion air interlocked with the appliances. The interlock is the load-bearing part: powered make-up air that is not interlocked so the appliance cannot fire without it is a hazard rather than a solution.

The adopted code. Your jurisdiction may adopt the International Fuel Gas Code rather than NFPA 54, and the combustion air provisions are parallel but not word-for-word identical. Size from the code the authority having jurisdiction has actually adopted, in the edition it adopted.

How to verify you got this right

Do not verify a combustion air correction by measuring the grille you just installed. Verify it with the two tests the building can fail.

First, put a manometer between the appliance space and outdoors and re-run the worst case exactly as the occupants live: every exhaust device on, interior doors in the position that most depressurises the space, and the clothes dryer actually running rather than merely switched on. Record the pressure. In this house it came back to about 2 Pa negative after the grilles were corrected and make-up air was provided for the range hood.

Second, re-run the spillage check at the draft hood after 5 minutes of main burner operation, in that same worst case, from a cold start. Spillage cleared in under 30 seconds, which is the result you are looking for: brief spillage during the first seconds of a cold start is normal while the vent warms, sustained spillage is not.

Write both results into the job record with the worst-case configuration you tested under. A grille size in the record proves nothing a year from now. A measured pressure and a spillage result under a stated building condition is something the next tech can actually re-test against when somebody adds another exhaust fan.

References

• NFPA 54 (ANSI Z223.1), National Fuel Gas Code, combustion air provisions, including the indoor-air standard method, the two-permanent-opening method, the louvre free-area assumptions and the draft hood spillage check; use the edition adopted locally
• International Fuel Gas Code, combustion air provisions, where adopted in place of NFPA 54
• 29 CFR 1910.1000 Table Z-1, general industry permissible exposure limit for carbon monoxide
• See related: How Combustion Actually Works; How Draft Actually Works