Low-Load Excess-Air Reduction Causes CO and Flame Instability
A 210 MW natural-gas-fired drum boiler is dispatched at 45% MCR for the evening. To keep NOx below the permit limit of 0.05 lb/MMBtu, the operator trimmed flue-gas O2 from 3.5% to 2.6% and closed the overfire air dampers by 15%. Within 30 minutes, stack CO rises from 35 ppm to 850 ppm, one of eight flame scanners flickers, furnace pressure oscillates ±0.6 in. w.c., and superheat attemperator spray drops to zero. The continuous emissions monitor is not yet out of range. The shift supervisor wants to hold until the wind picks up; the boiler is on a must-run contract. The unit has no CO catalyst. The burner management system has a low-O2 alarm at 2.0% and a trip at 1.5%.
1. What is the most likely physical explanation for the CO rise, scanner flicker, and furnace-pressure oscillation, and why is 2.6% O2 not evidence of safe combustion?
Read the worked answer
At 45% load the burner zone is already at its stability limit. Lowering O2 to 2.6% and closing overfire air reduced the local excess air at the burner throats; the burners are now running fuel-rich in the near-field, producing CO and unburned hydrocarbons even though the average stack O2 looks acceptable. The flickering scanner is consistent with a detached or unstable flame, and the furnace-pressure oscillation suggests localized puffs or delayed ignition. Average O2 is a mixing-weighted number; it cannot prove that each burner has enough air. The drop in attemperator spray also indicates the furnace heat-release pattern changed. The correct reading is that the unit is approaching incomplete combustion and a possible furnace explosion, not that it is “almost out of CO margin.” CO at 850 ppm with a rising trend and unstable flame is a trip-level precursor even if the BMS low-O2 alarm has not yet annunciated.
2. Give the immediate operator actions and the operating limits you would set for the next four hours. Include what you would not do.
Read the worked answer
Immediately raise flue-gas O2 back to the pre-trim 3.5% or higher, reopen the overfire air dampers to the commissioning position, and reduce load if CO continues to rise or a second scanner becomes unreliable. Bring in the combustion engineer and review the burner management system trends. I would set hard limits: CO above 500 ppm for more than 5 minutes, any scanner loss, or furnace pressure exceeding ±0.5 in. w.c. → reduce load and stabilize; CO above 1,000 ppm or a second scanner loss → controlled shutdown. Do not continue to chase NOx by lowering O2; do not defeat the low-O2 trip; do not rely on the CEMS not being out of range. Notify the grid operator of a possible derate. The four-hour plan should include a formal management-of-change review, because the operator has effectively changed the combustion setpoints outside the OEM curve.
3. What longer-term fix would you require before the next must-run season? Address tuning, instrumentation, and governance.
Read the worked answer
The fix is not a single setpoint. First, perform a burner-to-burner combustion tune at 40%, 60%, and 80% load with individual burner O2/CO measurements, not just stack O2. Second, verify scanner sighting, cooling air, and discrimination settings; a flickering scanner at low load may be a flame-signal issue as much as a flame-stability issue. Third, add or repair local CO monitoring at the furnace exit or economizer outlet if the plant only has stack CO, and trend it in the DCS. Fourth, revise the operating procedure to prohibit low-load O2 reduction below the OEM curve without a management-of-change and a combustion test. Finally, the NOx permit limit should be met by combustion tuning, overfire air optimization, or post-combustion control-not by running the burners fuel-rich. If the unit cannot make NOx at 45% load without unstable combustion, the correct answer is a derate or an emissions-control project, not an operating workaround.