High CO at High Fire After a Gas Heating Value Change
A 600-hp packaged watertube boiler fires natural gas through a parallel-position burner. At high fire the portable analyzer at the stack sample port reads 4.5 % O2, 9 % CO2, 520 °F, and 60 ppm CO. A permanently mounted probe at the furnace outlet reads 1.8 % O2 and 800 ppm CO. The forced draft fan runs at 98 % of full-load amps at high fire. The gas supplier has been blending in richer gas: billing shows 1,080 Btu/ft³ where the burner was commissioned on 1,000 Btu/ft³. Gas flow is set by a fixed orifice and volume regulation, so the burner still fires the same cubic feet per hour. The boiler has run this way for three weeks and there is a light haze at the stack on high fire.
1. Calculate excess air from both oxygen readings. Which reading should guide combustion decisions, and why do they differ?
Read the worked answer
Excess air approximation: %EA ≈ O2 / (20.9 − O2) × 100.
- Furnace-outlet probe: 1.8 / (20.9 − 1.8) × 100 ≈ 9.4 % excess air.
- Stack probe: 4.5 / (20.9 − 4.5) × 100 ≈ 27.4 % excess air.
Why they differ. An 18-point gap is far too large for instrument error. It is air in-leakage between the furnace exit and the stack sample point - a leaking expansion joint, economizer bypass damper, or casing penetration. The stack analyzer is diluting the sample with tramp air and is therefore telling you a comfortable lie.
Which governs. Combustion decisions must be made on gas leaving the furnace, before any leakage: the furnace-outlet probe at 9.4 % excess air. That is already at the bottom of the normal band for gas firing and, at 800 ppm CO, the burner is confirming incomplete combustion. Note the O2 formula assumes dry, CO-free gas; at 800 ppm CO the correction is small, but the true excess air is slightly lower than the calculation.
The 60 ppm CO at the stack is also not trustworthy - it is diluted by the same leakage.
2. Explain why the fuel change produced this condition and list the actions for the next shift.
Read the worked answer
Mechanism. The gas train meters by volume. At the same cubic feet per hour, 1,080 Btu/ft³ delivers about 8 % more heat input than the 1,000 Btu/ft³ the burner was commissioned on. That extra input needs extra combustion air, but the FD fan is already at 98 % amps and cannot supply it. The mixture goes rich, CO climbs, and the haze appears. The 520 °F stack temperature is partly the richer flame and partly the extra duty.
Actions, in order:
- Reduce the maximum firing rate until furnace-outlet CO is inside the target agreed with the burner manufacturer and the air permit (commonly well under 100–200 ppm at high fire). Record the O2/CO/temperature at each step.
- Put a documented high-fire limit in the control system so night shift cannot simply raise it back.
- Check the gas supply pressure at the burner manifold against nameplate while doing this - a pressure problem can masquerade as an air problem.
- Notify the gas supplier, obtain the Wobbe index and heating value, and have the burner re-commissioned with the correct orifice and linkage settings for the new gas.
- Correct the metering so input is compensated for heating value, rather than volume alone.
Do not chase the CO by opening the stack damper or 'trimming to the analyzer' at the stack.
3. The plant manager proposes raising the stack O2 target to 6 % so the stack analyzer 'reads right.' Respond, and state the correct basis for trim.
Read the worked answer
Reject the proposal. Raising the O2 target to 6 % at the stack would set the furnace to roughly 6 / (20.9 − 6) ≈ 40 % excess air if the leakage were ever sealed - a large stack loss and a change to the burner's approved firing envelope. Worse, it does nothing about the CO because the stack probe is not measuring the furnace.
Correct basis for trim:
- Trim from the furnace-outlet (or burner-mounted) probe, at each firing rate, using a step test across the range from low to high fire.
- Target the O2/CO band the burner manufacturer specifies for this gas and this furnace, cross-checked against the air permit limits for NOx and CO.
- Fix the leakage before trusting any stack reading: repair the expansion joint or casing leak, then re-baseline the stack analyzer so it agrees with the furnace probe within roughly 1 % O2.
- Log the readings each shift; a slow upward drift in furnace CO at constant O2 is the classic early warning of a fouled burner or falling gas pressure.