455°F Stack on a New 80% Furnace: What the Numbers Do and Don't Prove
An 80% AFUE, single-stage induced-draft furnace rated 90,000 Btu/h is being commissioned in a new townhouse at sea level. The rating plate specifies 3.5 in. w.c. manifold pressure and a 35-65°F temperature rise. After 15 minutes of steady operation the technician records: manifold 3.5 in. w.c.; return air 70°F; supply air 128°F; flue temperature 455°F taken 18 in. downstream of the flue collar; O2 9.2%; CO2 6.4%; CO 14 ppm as-read; flue draft -0.02 in. w.c. The analyzer was zeroed in outdoor air 40 minutes earlier and its battery icon shows one bar. The burner flames are blue with slight lifting at the ports, and the burners are clean. The closet has a louvered door, and the return duct is sealed to the furnace. The supervising technician, reviewing the printout, says: "Lean the gas valve out a little and bring the CO2 up and the stack temperature down before we leave."
1. Are the O2 and CO2 readings internally consistent? Show the arithmetic and state what the agreement (or lack of it) actually proves about the sample and the instrument.
Read the worked answer
Excess air is approximated as O2 / (20.9 - O2) x 100 = 9.2 / 11.7 x 100, about 79%. For natural gas at sea level the stoichiometric CO2 is roughly 11.7% dry, so the CO2 expected at that O2 is about 11.7 x (20.9 - 9.2) / 20.9 = 6.6%. The analyzer reports 6.4%; that is inside the tolerance you would expect from a clean sample.
The agreement means the sample train is not leaking or diluting. A cracked hose, a wet filter, a partially blocked probe or a probe touching the flue wall would pull O2 up and CO2 down and break the relationship. It does not validate the instrument. The one-bar battery and the 40-minute-old zero are real risks: recharge or swap the battery, re-zero in outdoor air, drain the water trap, and re-sample before any number goes on the invoice. It also does not tell you the combustion is acceptable - 79% excess air is roughly double the usual 30-50% for this type of burner, and that, not the CO2, is the finding worth chasing.
2. The stack is 455°F and the temperature rise is 58°F. What do these two numbers tell you, and what must be verified before anyone touches combustion?
Read the worked answer
The rise is inside the 35-65°F band but low-normal, while the stack is high. Using the common natural-gas approximation, flue loss is about (455 - 70) / 15, roughly 26%, so steady-state efficiency is in the mid-70s - well under what an 80% unit should show, and the approximation is generous at only 6.4% CO2, because the extra flue-gas mass carries more heat per cubic foot of gas burned.
Check in this order before adjusting anything: (1) airflow - filter, blower speed tap versus the plate, external static, coil condition. Low airflow raises stack temperature with no combustion fault at all, and a high-speed tap raises excess air and lowers the rise. (2) Gas input - clock the meter and compare with the 90,000 Btu/h rating; a wrong orifice or a mis-set regulator changes every reading downstream. (3) The probe and instrument - location, depth, water trap, fresh zero. (4) The burner box and inducer seals, and whether the inducer is the correct replacement part. Only after those are known good is the excess air itself a combustion finding rather than an airflow or measurement artifact.
3. Evaluate the supervisor's instruction to 'lean the gas valve out.' What is the correct action and why?
Read the worked answer
Refuse it, politely and on the record. On a fixed-orifice induced-draft burner the gas valve regulator sets manifold pressure, and manifold pressure sets input. Backing it down does not add air; it derates the appliance below its listed input, changes flame velocity and shape, can produce delayed ignition and higher CO, and shifts the temperature rise - all while making the CO2 look better on paper. It also takes the appliance outside its listing and can void the warranty.
The defensible path is: re-zero the analyzer with a fresh battery and confirm steady state (stable return temperature for 10 minutes); clock the meter to confirm 90,000 Btu/h; confirm airflow and static; then look for the real source of excess air - burner-box gasket, inducer gasket, a wrong inducer, missing or oversized orifice spuds, or a flue probe reading a dilution stream. If everything is per the manufacturer and the O2 is still 9.2%, call the manufacturer's technical line with model and serial before changing a setting. If they authorize adjustment, set the manifold to the plate value with a calibrated manometer at steady state, then re-clock and re-analyze. Document the readings you started with.