Industrial combustion systems employ various methods to control flow of fuel (primarily natural gas) and oxidizer (air, O2 or combination). The combustion control system must control the flow of oxidizer in the correct proportion to the fuel flow in order to maintain the desired furnace atmosphere chemistry (oxidizing, reducing or neutral).

     Fuel, combustion air and/or O2 flows are controlled by adjustable flow-control valves. These gases are supplied under pressure and are “pushed” into the furnace through the burner(s). Flames propagate through the furnace, delivering heat. The hot flue gases exit out the furnace flue, imposing a negative pressure “stack draft” on the furnace. This stack draft can cause significant quantities of infiltration air to be “pulled” through the furnace, depending on control of the furnace pressure, which is typically controlled by an adjustable flue damper. Infiltration air is pulled through cracks or gaps in the furnace refractory, door seals, burner mounts or other access ports. The more negative the furnace pressure, the greater the quantity of infiltration air. The result is that the infiltration (dilution) air will cool the furnace and reduce thermal efficiency, thereby increasing fuel costs. Infiltration air will also increase O2 content in the furnace atmosphere, which can be detrimental to many heating/melting process loads, such as aluminum melting or steel billet reheating.

     Overly positive furnace pressure can also be detrimental. Under positive furnace pressure, flames and/or hot furnace gases can squeeze out of furnace cracks, gaps or door seals (stingers), causing damage to these areas or causing unsafe conditions for personnel. In some processes, positive furnace pressure can cause smoke or fumes to escape outside of the exhaust-handling system, creating fugitive emissions, which can be dangerous and potentially in violation of environmental regulations.

     Because of these detrimental effects of positive furnace pressure, many furnaces err on the side of excess negative pressure out of practicality. For example, flue cross sections can be oversized and/or flue dampers maintained in wide-open positions without adjustment.

     In order to accurately control furnace pressure, and thereby accurately control the real overall oxidizer/fuel ratio and furnace atmosphere composition, the furnace flue damper should be controlled just as carefully as the fuel, air and O2 flow-control valves. In reality, controlling a furnace flue damper can be much more difficult than controlling a gas valve. In many cases, therefore, it is not controlled as carefully. Flue damper control can be especially important during low firing-rate conditions (holding or idling) and when employing oxyfuel burners because the reduced combustion product volumes can cause infiltration air to increase.

     To properly control any heating or melting furnace combustion system, the following parameters need to be known by the operator:

  1. What is the oxidizer/fuel ratio?
  2. Are the indicated flow meter and ratio data accurate?
  3. Is there significant infiltration air being pulled into the furnace? How much?
  4. What is the actual furnace chemistry (oxidizing, reducing, neutral)? How do I know? How can I be sure? Can it be quantified?

     Typically, oxidizer/fuel ratios in combustion control systems are set based on the assumption that natural gas is 100% CH4. This is the basis for the commonly employed 2:1 oxy/gas ratio and the 10:1 air/gas ratio (which includes 5% excess air). The actual natural gas composition can be different, however, depending on the source of natural gas or even the time of year, which may dictate adjustments to the air/gas or oxy/gas ratios.

     For fuels other than natural gas, such as oil or other liquid fuels including recycled waste fuels, the exact composition may not be well known.

     Also, flow meters for natural gas, other fuels, combustion air and O2 will typically exhibit some inaccuracy. Flow readings can be off by a small percentage or by as much as 10% or more. The accuracy depends on the type of meter, its calibration, maintenance and compensating factors such as temperature, pressure, dust level and humidity. It is especially challenging to accurately measure combustion air flow due to the larger volumes, lower supply pressures, higher dust loadings, variation in ambient humidity and temperature, and difficulty of providing long straight-length piping sections to improve flow meter accuracy.

     Considering the combined effects of fuel and oxidizer flow-meter inaccuracies, imperfectly known fuel composition and unknown furnace infiltration air level, the only way to accurately determine actual furnace atmosphere composition (i.e. actual overall oxidizer/fuel ratio inclusive of infiltration air) is by measuring it with gas analyzers.

     In many cases, one can determine visually if a flame is oxidizing or reducing based on flame color, shape, etc. But even with the best visual observation, one cannot quantify the actual flue-gas composition or the extent of oxidizing or reducing conditions. The only way to quantify the furnace atmosphere correctly is by measuring it.

     In this paper, furnace flue-gas composition measurements from several industrial melting furnaces are presented and discussed. A useful graphical technique (V curve) is employed to compare these measurements to calculated values and to better understand the combustion-control system performance, including calculation of furnace air-infiltration levels.


Definitions and Combustion Equations for Air/Oxy/Gas Combustion

G = natural gas flow rate in SCFH. Assume natural gas is 100% CH4. Use G = 1 as a basis.

A = actual air/gas ratio (combustion air plus infiltration air). A = Ac + Ai

     Where Ac = combustion air/gas ratio; and Ai = infiltration air/gas ratio


X = oxygen/gas ratio (from pure O2 i.e. purchased O2)

Perfect (stoichiometric) combustion equation: 1 CH4 + 2 O2 → 1 CO2 + 2 H2O

     Theoretically, 2 SCF (moles) O2 are required per 1 SCF (mole) of CH4 for perfect combustion. Note that for complete CH4 combustion, O2 is distributed equally between C and H.

     In real industrial combustion systems, the exact amount of total combustion O2 for a given amount of fuel is rarely provided. Depending on the furnace process and combustion controls, there is either an excess of O2 (oxidizing conditions) or a deficiency of O2 (reducing conditions).

     The term Lambda (L) has been developed to quantify the total amount of combustion O2 supplied:

  • Lambda = 1.0: perfect stoichiometric combustion (2:1 ratio of O2 to CH4)
  • Lambda > 1.0: oxidizing conditions (lambda = 1.05 means 5% oxidizing, or 5% excess O2)
  • Lambda < 1.0: reducing conditions (lambda = 0.95 means 5% reducing, or 5% deficiency of O2)

     For air/oxy/gas combustion, the term oxygen participation (P) is used to quantify the amount of combustion O2 provided by pure (purchased) O2 versus air. Air is assumed to be 21% O2 and 79% N2.

     Relating the oxy/gas ratio X and the total air/gas ratio A to lambda (L) and O2 participation (P):

                P = X / (0.21A + X) and L = (0.21A + X)/2

                or, X = 2PL and A = (2L – X)/0.21

     For example, with air/gas combustion, X = 0 and P = 0. With 100% oxy/gas combustion (oxy/fuel burner), P = 1 and A = 0 with respect to combustion air supplied by a blower Ac (A can be > 0 and P < 1 considering infiltration air Ai). For air/oxy/gas combustion, P will be between 0 and 1.


For oxidizing conditions (excess O2 is supplied):

1 CH4 + X O2 + A (0.21 O2 + 0.79 N2) → 1 CO2 + 2 H2O + 0.79A N2 + (0.21A + X – 2) O2


For reducing conditions (not enough O2 is supplied for complete combustion):

Incomplete products of combustion will include CO and H2 in addition to CO2 and H2O:

1 CH4 + X O2 + A (0.21O2 + 0.79N2) → d CO2 + m CO + w H2O + h H2 + 0.79A N2 + 0 O2


     In the Appendix, equations for flue-gas species concentrations (dry basis) are outlined, for both oxidizing and reducing conditions. Equations for calculating furnace infiltration air are also shown.


Graphical Presentation of Combustion Conditions: V Curve

Using the relations outlined above and in the Appendix, flue-gas compositions (furnace atmosphere chemistry) can be calculated, for various values of lambda (L) and O2 participation (P). It is useful to look at a graph showing flue-gas composition versus lambda for a given value of O2 participation.

     Figure 1 shows flue-gas compositions are calculated as a function of lambda for a fixed O2 participation of 50% (P = 0.50). For oxidizing conditions (L > 1), there will be excess O2 in the furnace atmosphere, with zero CO and zero H2. With reducing conditions (L < 1), CO and H2 are present with zero O2. At perfect “stoichiometric” or neutral conditions, CO2 concentration reaches a peak (sometimes referred to as ultimate CO2), and O2, CO and H2 are theoretically zero (in reality, there can be some CO and H2 at this point, owing to imperfect fuel/oxidizer mixing).

     For any O2 participation P (0 <P < 1), a similar “V-shaped” curve can be generated. The neutral point will be at the bottom of the “V,” at calculated lambda = 1.0.

      It is important to remember that these calculated flue-gas compositions are for the products of combustion (POCs) only. This assumes no chemical interaction between the furnace load and the combustion atmosphere. In reality, depending on the heating/melting process and the furnace load, there can be chemical reactions between the combustion atmosphere and the load. Understanding this, it is still useful to consider the furnace atmosphere created by the products of combustion.


Measured Flue Gas Compositions: Field Data from Several Melting Furnaces

Flue-gas composition (O2, CO2, CO, H2) measurements for several different melting furnaces are presented and discussed below. In three of these cases, lambda was intentionally varied over a wide range in order to generate a V curve similar to Figure 1. The V curve provides an informative graphical illustration for comparing measured flue-gas compositions to calculated (predicted) values. In all three chosen examples, the measured data closely follow the V shape of the calculated data.

     In other cases, a full V curve “lambda traverse” was not generated, but a plot of O2 versus lambda for even a few data points provides a useful illustration of combustion conditions.

     The types of furnaces analyzed below include: closed-barrel, rotary aluminum melt furnace, well-charged aluminum reverb melt furnace, copper reverb batch melt furnace and a glass reverb melt furnace. In all cases, flue-gas compositions were measured with either an empty furnace or during conditions when chemical interaction between the furnace load and the combustion atmosphere is negligible (flat-bath melting or holding for aluminum and glass and flat-bath holding for copper).

     Examples of natural gas-fired air/fuel, oxy/fuel and air/oxy/fuel systems are included, and in one air/oxy/fuel example the furnace was fired with recycled (waste) oil.

     Flue-gas compositions are reported as dry basis (H2O removed) because gas analyzers utilize dry flue gas samples (H2O is removed from the flue gas sample via chiller/condenser unit).

     These example furnace “case studies” illustrate how flue-gas analysis can provide increased understanding of combustion conditions and control-system operation to improve combustion-system performance. In some of these examples the flue-gas analysis helped to diagnose and correct problems with the combustion-control system.


Closed-Barrel, Rotary Aluminum Melting Furnaces: Air/Oxy/Gas and Oxy/Gas

In a closed-barrel rotary furnace, the burner fires into the horizontal barrel-shaped furnace at the open barrel end. The burner is mounted to a door structure, which closes this end of the furnace and has a flue opening positioned above the burner. The burner flue gases turn around inside the furnace and exit through this flue opening. This is geometrically similar to a horizontal ladle heater in a steel mill.

     For one example – air/oxy/gas rotary aluminum melter – flue-gas composition was measured at high fire conditions (9 MMBTU/hour with 50% O2 participation) over a range of lambda values while firing into the hot, empty furnace. The goal was to establish a “calibration V curve” similar to Figure 1 and to determine how closely the measured data correlated with calculations. The results, shown in Figure 2, indicate that for this furnace at these firing conditions there was very close agreement between measured and calculated flue-gas composition.

     Figure 2 indicates that the selected set-point lambda will provide the expected combustion atmosphere for this furnace. In this example, the operator could proceed with confidence that the chosen combustion settings would provide the expected furnace atmosphere at high-fire conditions. Figure 2 indicates that the gas, air and O2 flow meters and flow controls are working as expected and the assumption of 100% CH4 composition for natural gas is valid.

     At the high-fire condition, it appeared that the furnace (closed barrel) was under positive pressure. And owing to the close agreement between measured and calculated flue-gas composition, it seemed reasonable to assume that there is no additional furnace infiltration air at high fire. Indeed, the calculated infiltration air level, utilizing the equations in the Appendix, is very low (between zero and about 2,000 SCFH) for the three oxidizing data points considered. Air of 2,000 SCFH corresponds to about 20% of the gas flow or about 4% of the combustion air flow. Therefore, it can be concluded that within the accuracy of these various measurements there is essentially zero air infiltration at this high firing rate.

     The burner was then changed to low fire (2 MMBTU/hour or 2,000 SCFH natural gas) at lambda = 0.95 and O2 P = 0.50. Measured flue-gas composition was 0% CO and H2, with 3% O2 and 16% CO2, indicating slightly oxidizing conditions even though lambda was set for 0.95. This suggested that there was some infiltration air being pulled into the furnace at low fire. Using measured O2 and the measured X oxy/gas ratio, calculated infiltration air was 2,602 SCFH, which in this case is 29% of combustion air at this low firing rate or 1.3X the gas flow. The effective lambda, including the infiltration air, is 1.09. This suggests that to maintain reducing conditions at low fire, lambda should be reduced to compensate for the additional infiltration air. The calculated value for the required lambda, to provide neutral atmosphere (zero [O2]) at low fire is 0.895.

     This observed increase in infiltration air when decreasing firing rate from high to low is very typical for any fuel-fired furnace. When decreasing firing rate, burner flue-gas volume (POC) decreases. Because there is a fairly constant “flue stack draft” pulling on the furnace, air infiltration rate naturally increases. This is especially true when there is no adjustable flue damper.

     Flue-gas composition data from two other closed-barrel rotary aluminum melting furnaces were also analyzed. These data are summarized in Table 1. One furnace was 100% oxy/gas, where the first two furnaces are air/oxy/gas. In all cases, as firing rate is decreased, the amount of infiltration air relative to POC flow is high enough to create increasingly oxidizing conditions (the effective lambda is higher than the lambda burner setting). So, when firing rate is decreased, it is necessary to reduce lambda in order to maintain a neutral atmosphere (prevent oxidizing conditions) due to the increased relative effect of infiltration air.

     As shown in Table 1, it is common for 100% oxy/gas burners to utilize a set-point lambda of less than 1.0, recognizing that the balance of combustion O2 requirement can often be met via infiltration air.

     Unlike reverb melt furnaces, with the closed-barrel rotary furnace, one cannot determine how much infiltration air is pulled through the furnace barrel versus pulled straight up and out the flue since the flue-gas measurements are taken in the flue stack above the burner.

     Therefore, flue-gas analysis for rotary furnaces can beneficially provide:

  • Calibration of the combustion system settings to actual furnace atmosphere
  • Determination of required reduction in lambda to maintain neutral conditions at low fire
  • Estimation of furnace infiltration air level
  • Aid in adjusting flue damper position (if adjustable)


Well-Charged Aluminum Reverb Melter with 100% Air/Gas Burners

Flue-gas composition was measured on a well-charged, flat-bath aluminum reverb melt furnace. This furnace utilized a conventional air/gas burner system (natural gas-fired) with two burners, employing a mechanical ratio regulator to control air/gas ratio. This system did not include a permanent combustion air flow meter. Supplemental O2 was not utilized.

     Data and calculations from high fire (14 MMBTU/hr) and low fire (3 MMBTU/hr) are shown in Table 2.

     At high fire, flue-gas composition measurements indicated slightly reducing conditions (zero O2 with 1-3% CO). This indicates that there was not quite enough combustion air to fully combust the fuel. However, these slightly reducing conditions are usually desirable in aluminum reverbs in order to minimize the potential for aluminum oxidation.

     At low fire, the very high O2 along with the low CO2 suggested that there was a large excess of air. It was observed that the combustion air pressure at the burner pressure tap was exactly the same as during high fire. This suggests that the air butterfly valve linkage was not turning down for low fire.

     At both high fire and low fire, the furnace pressure appeared significantly positive, and the draft gauge indicated positive pressure. This implies that there is little or no additional infiltration air supplementing the combustion air from the blower.

     As shown in Table 2, combustion air flow in SCFH was calculated by two methods: using measured flue-gas CO2 concentration and measured flue gas O2 concentration. In this example, these two methods show good agreement. Note that in this case, since there was no combustion-air flow meter, this calculation of air flow calculates total air flow (combustion air plus infiltration air). Owing to the observed positive furnace pressure, however, one can conclude that infiltration air is essentially zero and all combustion air is from the blower.

     The calculated combustion air flow is essentially the same for both high fire and low fire, supporting the observation that the air butterfly valve position had not changed between high and low fire. The calculated air/gas ratio of 8.5 to 9 at high fire indicates reducing conditions, while the calculated 40:1 air/gas ratio at low fire is extremely high.

      In next month’s online exclusive, we continue our discussion and look at more examples, including aluminum and glass reverb melt furnaces.


Appendix: Flue Gas Composition Equations, for Oxidizing and Reducing Conditions

Note that for ideal gases, SCF (standard volume) is proportional to moles. The ideal gas assumption is very appropriate, for calculating gas compositions for products of combustion at atmospheric pressure (low pressure and/or high temperature).

For oxidizing conditions (excess O2 is supplied):

1 CH4 + X O2 + A (0.21 O2 + 0.79 N2) → 1 CO2 + 2 H2O + 0.79A N2 + (0.21A + X – 2) O2

Total POC (products of combustion) = A + X + 1

Gas analyzers operate on a dry sample basis (water vapor removed), so dry flue gas volume = A + X -1

Dry flue gas concentrations, expressed in terms of X and A, are:

  • [CO2] = 1 / (A + X -1)
  • [O2] = (X + 0.21A -2)/(A + X -1)
  • [N2] = 0.79A / (A + X – 1) = 1 – [O2] – [CO2]
  • [CO] = [H2] = 0

Oxidizing conditions are normally encountered in most industrial combustion processes, as it is normally desirable to ensure that sufficient total O2 is supplied for complete combustion of fuel.

One useful application of flue-gas analysis is to utilize the measured flue gas compositions to calculate the amount of (extra) infiltration air being pulled through the furnace in addition to the combustion air provided by a blower. Infiltration air amounts can be more significant at lower firing rates, or with oxy/fuel or air/oxy/fuel burners, as the decreased burner flue gas volume (POC volume) can cause the furnace to draw in greater quantities of infiltration air, especially without flue damper adjustment or control (furnace pressure control).

With oxidizing conditions, to calculate infiltration air, it is often most accurate to utilize the measured dry (O2) flue gas concentration and the measured oxy/gas ratio X, then back-calculate the actual total air/gas ratio A (since in some cases measured [O2] can be more accurate than [CO2]):

                A = [X([O2] -1) – [O2] + 2] / (0.21 – [O2]) = total A (combustion air plus infiltration air)

Or if there is reason to suspect the accuracy of the measured oxy/gas ratio X, then in some cases it can be more accurate to utilize measured [CO2] and [O2] values, and use the [N2] equation to calculate A:

                A = (1 – [CO2] – [O2]) / (0.79 * [CO2])


For reducing conditions (not enough O2 is supplied for complete combustion):

Incomplete products of combustion will include CO and H2 in addition to CO2 and H2O:

1 CH4 + X O2 + A (0.21O2 + 0.79N2) → d CO2 + m CO + w H2O + h H2 + 0.79A N2 + 0 O2

Assuming that O2 is again distributed equally between C and H (half goes to CO2 and CO, and half goes to H2O), then equal amounts of CO and H2 are produced (m = h). By C, H and O balance, it follows that:

m = h = 2(1 – L); and d = (2L – 1); and w = 2L

Total POC = 3 + 0.79A; dry flue gas volume = 1 + h + 0.79A = 3 -2L + 0.79A


Dry flue gas compositions, expressed in terms of L and A, are:

[CO] = [H2] = [2(1 – L)] / (3 – 2L +0.79A)

[CO2] = (2L -1) / (3 – 2L + 0.79A)

[N2] = 0.79A / (3 – 2L +0.79A) = 1 – [CO] – [H2] – [CO2]

[O2] = 0


Stewart Jepson, Air Liquide - Combustion Specialist, 800-820-2522