CO in the Supply Air on an Overfired Furnace: Correct the Input First
An 18-year-old, 100,000 Btu/h, 80% upflow furnace is being checked on a seasonal tune-up. The rating plate lists 3.5 in. w.c. manifold and a 40-70°F temperature rise. Clocking the 2 ft³ meter dial gives one revolution in 60 seconds. Manifold pressure at the valve outlet is 4.2 in. w.c. After 20 minutes of operation: return 68°F, supply 132°F, flue 470°F, O2 4.1%, CO2 8.8%, flue CO 210 ppm as-read, draft -0.03 in. w.c. Ambient CO in the basement is 1 ppm; a handheld monitor held at the nearest supply register reads 8 ppm after 5 minutes. Burners show light rust but no visible flame roll-out, and the blower wheel is heavily loaded with dust. The helper says: "CO in the supply means the heat exchanger is cracked - we have to red-tag it today." The homeowner has a working CO alarm and reports no symptoms.
1. Using the clocked gas rate and the manifold pressure, determine whether the furnace is overfired, and state what to correct first.
Read the worked answer
One revolution of the 2 ft³ dial in 60 seconds is 2 ft³/min, or 120 ft³/h. At roughly 1,030 Btu/ft³ that is about 123,600 Btu/h against a 100,000 Btu/h rating - approximately 24% overfired. Manifold pressure explains only part of it: flow scales with the square root of pressure, so 4.2 vs 3.5 in. w.c. is sqrt(4.2/3.5) = 1.095, about 9.5%.
The rest has to come from the orifices or from a clocking error. Before condemning anything: confirm the clocking over at least two full dial revolutions or three minutes with no other appliance drawing, and confirm the local heating value with the utility. Then set the manifold to 3.5 in. w.c. with a calibrated manometer, referencing the burner box if the valve is a negative-pressure type, re-clock, and if it is still over, pull the orifices and check the drill size against the plate. Correct the input before interpreting any combustion number, because 24% overfire alone will drive stack temperature up, excess air down, and CO up.
2. Interpret the flue readings and the 8 ppm at the register. What does the analyzer prove, and what does it not prove?
Read the worked answer
Air-free CO is 210 x 20.9 / (20.9 - 4.1) = about 261 ppm - high for a clean natural-gas furnace, which should be well under 100 ppm air-free. O2 4.1% gives about 24% excess air, and the CO2 expected at that O2 is 11.7 x (20.9 - 4.1) / 20.9 = 9.4%, close to the measured 8.8%, so the sample is sound. Low excess air plus high CO plus a 470°F stack plus a loaded blower wheel is a picture of overfiring and restricted airflow, not proof of a failed heat exchanger.
The 8 ppm at one register versus 1 ppm ambient is above ambient and cannot be waved off, but five minutes at one register is not a test. The monitor may be drifting, the duct may be pulling attic or garage air, and a small rise can come from a source other than the furnace. The analyzer cannot prove a heat exchanger is intact - it can only tell you combustion is poor. The defensible sequence is to correct input and airflow, then repeat a steady-state supply-plenum CO comparison against ambient before drawing any conclusion.
3. What do you actually do on this call, and what would change your decision to shut the furnace down?
Read the worked answer
Today: set the manifold to 3.5 in. w.c., clean the blower wheel and the burners, verify the temperature rise and external static, re-clock the meter, and re-run the combustion analysis. Then re-test the airstream properly - a calibrated monitor in the supply plenum at steady state, compared with ambient, repeated at the farthest register. Document pre- and post-repair readings, the clocked input, and the customer conversation.
I would not red-tag on the 8 ppm alone, but I would not clear the heat exchanger with an analyzer either. If the supply rise above ambient persists after the input and airflow are corrected, if the flame distorts when the blower starts, if there is visible cracking or a failed inspection port, or if I cannot inspect the exchanger, the furnace goes off and is tagged - with the manufacturer's inspection procedure done by camera at the cell inlet and outlet and the vestibule seams before any return to service. The customer gets a plain explanation: the unit was overfired and starved of air, that is corrected, the CO readings are the evidence, and the CO alarms should stay in place on every level.