Gas analysis for O₂, CO₂ and dry molecular weight
EPA Methods 3, 3A and 3B, 40 CFR Part 60, Appendix A-2
Method 3 finds out what the stack gas is made of. Oxygen and carbon dioxide set how heavy the gas is, how much extra air the burner is using, and how results are corrected to a common basis.
- Answers
- How much CO₂ and O₂ is in the gas, and what does a mole of dry gas weigh?
- Equipment
- Probe, sampling bag and leak-free pump, Orsat or Fyrite analyzer, or instrumental O₂ and CO₂ analyzers
- Produces
- %CO₂ and %O₂ (dry), dry molecular weight M_d, excess air
- Feeds into
- Method 2 velocity, Method 5 isokinetic setup, oxygen corrections and F-factor emission rates
Why gas composition matters
Combustion takes oxygen out of the air and puts carbon dioxide and water vapor in. That changes the gas in three ways a stack test has to account for:
- Density. CO₂ (44 lb/lb-mol) is much heavier than the nitrogen (28) it displaces. Method 2 needs the molecular weight to turn Δp into velocity, and Method 5 needs it to set the isokinetic sampling rate.
- Dilution. A boiler can lower its concentration of a pollutant just by adding more air. Correcting results to a reference oxygen level, such as 7% or 15% O₂, removes that loophole.
- Fuel-based rates. Emission rates in pounds per million Btu use O₂ or CO₂ with fuel F-factors (EPA Method 19).
All Method 3 results are on a dry basis: the water has been removed before the gas is analyzed. Moisture comes separately from Method 4.
Dry molecular weight
The dry molecular weight is a weighted average of the gases in the sample. Each gas contributes its molecular weight times its volume fraction. Nitrogen is found by difference and stands in for everything else, including the 1% of argon in air.
Dry molecular weight
- 0.440,\ 0.320,\ 0.280
- molecular weights of CO₂ (44), O₂ (32), and N₂ and CO (28), divided by 100
- \%N_2
- 100 − %CO₂ − %O₂ − %CO, all by volume on a dry basis
Composition mixer
Start from a typical source, then move the sliders.
The presets are built from EPA Method 19 F-factors, so CO₂ and O₂ stay consistent for each fuel. Fo is a quick plausibility check: each fuel burns to a characteristic ratio, about 1.75 for natural gas, 1.35 for fuel oil and 1.14 for bituminous coal. A value far from the fuel’s ratio points to an analyzer or absorption problem. Diluting the gas with air does not change Fo, so it will not reveal a leak.
Collecting the sample
Method 3 allows three ways to collect gas, depending on what the result is for:
- Single-point grab sample. A quick draw from one point near the center of the stack. Good enough for molecular weight on a well-mixed stack.
- Single-point integrated sample. Gas pumped steadily into a flexible bag over the whole run, so the result is a time average.
- Multipoint integrated sample. The same, but the probe moves through the traverse points. This is the usual choice when the O₂ or CO₂ result will be used to correct an emission rate.
An integrated train has a probe, a condenser to knock out moisture, a leak-free pump, a rate meter to hold the flow constant, and a bag that does not react with the gas. The whole train, bag included, is leak checked before use, because a leak pulls in room air and drags the result toward 20.9% O₂.
Integrated bag sampling train
Gas flows from the stack, left, into the bag, right.
Orsat analysis
The Orsat analyzer measures gases by making them disappear. A burette draws exactly 100 mL of sample. The sample is pushed into a pipette of chemical solution that absorbs one gas, then drawn back, and the burette shows how much volume is gone. Each pipette removes a different gas:
- Potassium hydroxide (KOH) absorbs CO₂.
- Alkaline pyrogallol absorbs O₂. It also absorbs CO₂, which is why CO₂ has to come out first.
- Cuprous chloride absorbs CO. It also takes up O₂, so oxygen has to be removed before it.
The gas is passed back and forth through each pipette until two readings in a row agree. What is left after the last pipette is counted as nitrogen. A Fyrite analyzer works on the same absorption principle for a single gas and is quicker, but less precise. Method 3 allows it for molecular weight, while results used to correct emission rates or calculate excess air need the more exacting procedures of Method 3B or the analyzers of Method 3A.
Orsat analyzer
Draw a sample, then pass it through the pipettes. Try them out of order.
Method 3A: continuous analyzers
Most modern tests measure O₂ and CO₂ with continuous analyzers under Method 3A instead of an Orsat. Oxygen is strongly attracted to a magnetic field, which paramagnetic O₂ analyzers exploit; others use electrochemical or zirconia cells. CO₂ absorbs a specific band of infrared light, so it is measured by nondispersive infrared (NDIR).
The gas passes through a conditioning system that chills it and drains off the water, so the analyzers read on a dry basis, the same as an Orsat. Method 3A does not set its own quality checks; it follows the calibration error, system bias, and drift procedures of Method 7E. Those are covered in the instrumental analyzers lesson.
Excess air and oxygen corrections
Burners run with more air than the fuel strictly needs, so that every bit of fuel finds oxygen. The extra is called excess air. Method 3B calculates it from the Orsat results:
Percent excess air
- 0.264
- the ratio of O₂ to N₂ in air (20.9 ÷ 79.1)
- 0.5\,\%CO
- oxygen that would still be used to burn the CO to CO₂
Regulations then often express a concentration at a fixed oxygen level, so that diluting the exhaust with air does not make it look cleaner. Common reference levels are 3% O₂ for boilers, 7% O₂ for incinerators, and 15% O₂ for combustion turbines. Your permit or rule sets which one applies.
Correcting a concentration to a reference oxygen level
- C
- measured concentration, dry basis (for example ppmvd)
- X
- reference O₂ level from the rule, percent
- \%O_2
- measured O₂ during the same run, dry basis
Correcting to a reference O₂
Watch what extra air does to the raw reading and the corrected one.
In this model the flame makes 100 ppmvd of NOₓ at stoichiometric conditions (0% O₂) and air dilutes it. The corrected value stays put no matter how much air is added, which is the whole point of the correction.
What a wrong molecular weight costs
Velocity depends on 1/\sqrt{M_s}, so a molecular weight error is halved on the way to the flow rate. That is small, but it is a bias, not random noise, and it applies to every run. It is also why Method 2 only lets you assume 29.0 for a source that emits essentially air.
Assumed versus measured molecular weight
Set the real gas, then pick what you assumed.
Acceptance criteria
Methods 3, 3A and 3B at a glance
- Leak check the sampling train and the bag before use.
- Orsat or Fyrite for molecular weight (Method 3): repeat the analysis until three results for Md agree within 0.3 lb/lb-mol, and average them.
- Results used for emission rate corrections or excess air need Method 3B (Orsat with tighter agreement between repeat analyses) or Method 3A analyzers; Fyrite is not acceptable there.
- Method 3A analyzers follow Method 7E checks: calibration error within 2% of span, system bias within 5% of span, and drift within 3% of span.
- An assumed Md of 29.0 is allowed only for sources emitting essentially air.
See the current method text for the exact agreement limits in Method 3B.
Field mistakes that cost runs
Absorbing out of order
Pyrogallol takes CO₂ as well as O₂. Pass the sample through it first and all the CO₂ gets reported as oxygen. Always run KOH, then pyrogallol, then cuprous chloride.
Air leaking into the bag
A leak dilutes the sample with room air: O₂ reads high, CO₂ reads low, and an oxygen correction makes the corrected emissions look worse than they are. Compare the bag with the continuous analyzers or the expected O₂ for the process; Fo will not help here, because air dilution leaves it unchanged.
Tired reagents
Absorbing solutions lose capacity with use. If readings keep creeping with every pass, the solution is spent. Checking the analyzer on ambient air, which should read close to 20.9% O₂, catches a weak pyrogallol pipette.