Automated combustion processes are all around us, from residential appliances and HVAC systems to large industrial boilers, fired heaters, and power plants. For all the differences in scale and purpose, the common element that all try to achieve is high efficiency through effective combustion control. Fuel cost is a major factor, and hence the emphasis on efficiency, but safety and emission considerations are also top of mind, particularly the latter for industrial applications. The type of combustion discussed in this article is a chemical reaction between the fuel and oxygen (O 2 ) and is therefore subject to basic stoichiometric factors. The correct number of O 2 molecules must be available to react with the corresponding number of fuel molecules.
As a practical matter, most combustors use atmospheric air, with air flow measured to control the O 2 supply. Air flow imbalance in either direction is problematic. If there is insufficient air (below the stoichiometric requirement, or fuel-rich combustion), unburned fuel goes out the stack. This wastes fuel, creates emissions and hazardous air pollutants. It also creates a potential safety issue should enough fuel subsequently mix with O 2 and ignite.
To further complicate matters, in the real world, combustion is rarely one hundred percent complete. There is typically some amount of unburned fuel in the flue gas, although trace amounts won’t have a material effect on total versus excess O 2 levels. However, significant levels of unburned fueled are likely at some point in a facility’s operating life and is more common than realized. This is unavoidable even in the most efficient burners. More on what this means in a moment.
If there is too much air (above the stoichiometric requirement, resulting in fuel-lean combustion), efficiency is reduced due to energy wasted heating the unnecessary volume of air. This is inescapable to some extent since approximately 80% of air is nitrogen, but excess air is less problematic for efficiency and safer for operation, although nitrogen oxides (NOx) emissions can increase with increasing excess air. For most combustors there is an ideal excess air to achieve good combustion, low emissions and high efficiency. Excess air and excess fuel both reduce efficiency, but excess air doesn’t reduce efficiency as much as the same volume of excess fuel.
Analyzing combustion
Anyone who has worked with an old-fashioned gas stove or heater can see the mixing process working by adjusting the burner air intake to achieve a perfectly blue flame. But the question emerges, what is the most practical way to optimize large-scale combustion for safety, efficiency and emissions? The most common answer is to measure and control the amount of O 2 remaining in the flue gas exhaust, but what is ideal? As just mentioned, combustion is often not perfectly complete, so some unburned fuel and O 2 go out the stack, even if the air and fuel mixture going into the burner is just right.
The area of concern is the amount of O 2 in excess of what is required to burn the amount of fuel, but looking at total O 2 content in the flue-gas stream can be deceiving if operators don’t fully understand what the measurement represents. The challenge is to determine how much of the O 2 in the flue gas is in excess of the stoichiometric amount. Operators normally want some amount of excess O 2 because it is undesirable to reduce air flow below the stoichiometric amount (Figure 1), but the exact amount depends on the fuel and combustion system. Erring on the fuel-lean side in most situations is more desirable than running fuel-rich. 
Natural gas fired equipment can run with as little as 1% to 2% excess O 2 since fuel and air can mix thoroughly.
When combustion is more difficult, the excess amount may need to be higher. For example, a pulverized-coal-fired boiler will run around 3 to 4% excess O 2 , and a municipal solid waste boiler will be even higher, perhaps 5 to 7% excess O2 . Most gas- and oil-fired installations will be well below 3% excess O 2 .
Controlling combustion
For industrial installations, there is a wide range of control strategies. At minimum, there will be an instrument monitoring fuel flow. Air flow will be metered, or at least controlled, to correspond to fuel flow. This type of scheme can be implemented using some formula (air volume per unit of fuel) for a rough calculation, but variability in oxygen demand from different fuel sources, and precision of fuel and air flow measurements, results in the necessity to monitor the actual O 2 content of the flue gas as well. There are two techniques commonly applied for flue gas O 2 measurement: a tunable-diode laser (TDL) analyzer and a zirconia sensor analyzer.
A TDL analyzer uses two sensing components (Figure 2) , a laser source and detector. 
When used in an extractive gas sampling analyzer, the laser emits a beam of light through a cell containing the process gas and then on to a detector. The light emitted is at a specific wavelength which is absorbed by O 2 molecules present in the process gas, attenuating the light intensity. This attenuation in light is measured by the detector and is directly proportional to the concentration of O2 molecules present. The analyzer then uses a mathematical model to determine the O 2 percentage based on the degree of attenuation.
A zirconia sensor analyzer (Figure 3) is placed in a duct in the flue-gas stream. Gases from the duct diffuse across the filter media into the diffusion chamber based on gas partial pressure equalization mechanisms. With the Rosemount™ sensor analyzer shown below, gases then diffuse through the neck of the O 2 sensor cell, which is packed with platinum coated beads, to the zirconia sensor media. The entire cell is embedded in a heater to maintain the zirconia media at a temperature of 736 C. 
The two approaches may seem similar at first glance, but their measurements have a subtle but critical difference.
A TDL analyzer measures total O 2 content in the flue gas, which means its reading shows an undifferentiated mix of left-over stoichiometric O 2 and excess O 2 . When there is unburned fuel present, there is no way to determine how much is actually excess and how much is leftover O 2 , so operators have no firm measurement to go by. If combustion becomes poor and unburned fuel increases, operators may mistakenly misinterpret the total O 2 content as excess O 2 and acting on that belief, they can drive the flue gas mix even richer. This risk can be corrected by adding a second analyzer designed to detect and quantify the unburned fuel components.
When unburned fuel is detected, actions can be taken to avoid reducing combustion air flow even though total O 2 readings are high, however this requires two measurements, total O 2 and combustibles. On the other hand, the Rosemount proprietary zirconia sensor has a particular characteristic which avoids this problem. Flue gas, including both leftover O 2 and any unburned fuel, flows into the sensor. Given the high temperature and high surface area of the platinum beads (Figure 4), combustion of any unburned fuel and leftover O 2 is completed through catalytic-enhanced oxidation, much like the process used in car exhaust catalytic converters to reduce CO and hydrocarbon emissions.
Since all combustion is completed in the sensor, any O 2 remaining in the sensor represents excess O 2, which is different than the total O 2 in the flue gas if unburned fuel is present. Because excess O 2 is directly related to excess air, operators can control air flow in real-time to maintain the ideal amount of excess O 2 .
Improving efficiency, emissions and safety
Controls for boilers, fired heaters, and other combustion processes are built to regulate combustion air to ensure operation with adequate, but not too much, excess air. Zirconium sensors measure excess O 2 , not total O 2 , which is directly related to excess air. Other technologies, such as TDL, measure total O 2 , which does not provide a direct measure of excess air when unburned fuel is present. Total O 2 is only equal to excess O 2 when combustion is 100% complete and no unburned fuel exists. While an operation is never intended to operate with unburned fuel, the risk remains and happens to some extent in virtually all applications.
When using a TDL analyzer, its total O 2 measurement can therefore cause operators or automated controllers to drive to unsafe fuel-rich conditions. A simple and durable zirconium sensor addresses this issue by measuring only excess O 2 , allowing operators to control combustion processes while maintaining safe and optimum conditions.