The first portion of this article was an online exclusive in June/July. We pick up that discussion with additional industrial examples illustrating the work on flue-gas analysis.
Well-Charged Aluminum Reverb Melter with 100% Oxy/Gas Burners
In the next example, if the low-fire air/gas ratio were to be reduced from 40:1 to 10:1, the estimated natural gas savings would be $22,000/year, assuming the furnace spends 30% of the time on low fire (at $3.50/MMBTU natural gas price). Additionally, an improvement in aluminum yield (reduced main-hearth dross formation) was expected by reducing the furnace atmosphere O2 content during low-fire holding conditions.
With a correctly functioning air/gas ratio control, it is predicted that at low fire (without flue-damper adjustment) there would be infiltration air. Flue-gas analysis could be utilized to calculate furnace infiltration air and to help adjust the damper position at low fire.
In this example, flue-gas analysis provided the following benefits:
- Determines and quantifies the actual furnace atmosphere (oxidizing vs. reducing) at both high- and low-fire conditions
Diagnoses the problem with the combustion-air control valve at low fire
Flue-gas compositions were measured from a well-charged aluminum reverb melt furnace fired with two 100% oxy/gas burners. With 100% oxy burners, as compared to air/gas burners, it is important to consider air infiltration and to pay attention to flue damper adjustment, since the POC volume from oxy/gas burners is considerably lower than air/gas (2:1 oxy/gas ratio vs. 10:1 air/gas ratio), especially at low firing rates.
Here, the flue gas composition was utilized to calculate furnace air infiltration in SCFH for high-, medium- and low-fire conditions. The flue damper remained in a fixed position (open) for all firing rates.
Measured flue-gas composition and calculated air infiltration rates are shown in Table 3. Measured (CO) was zero in all cases, indicative of the oxidizing conditions (excess O2).
The relatively high flue gas (O2) concentrations shown in Table 3 indicate a high level of excess air. Infiltration air values are calculated via the equations in the Appendix (found in Part 1). In this case, it was more accurate to utilize the measured CO2 and O2 values to calculate Ai.
It is interesting to see how infiltration air increases as firing rate is reduced. Looking at the burner POC and total POC values, it appears that the furnace stack “draft” maintains a constant value of total POC volume (inclusive of infiltration air), by increasing infiltration air as firing rate (burner POC) is reduced.
Automatic Flue-Damper Control
This next example illustrates the advantage of an automatic flue-damper control system (furnace pressure control). Alternately, the flue-damper position could be controlled in a feed-forward fashion based on firing rate. Damper control would reduce infiltration air at the lower firing rates.
It is also important to note the very high 26.8% O2 flue-gas concentration at low fire. This indicates oxygen-enriched conditions (O2 content above 21%) in the furnace atmosphere and flue gas. Upon further investigation, the furnace operators indicated that one of the burners had been damaged or misadjusted, and an excessively high oxy/gas ratio was required, at low fire, to keep the burner lit (positive UV signal). The flue-gas analysis helped to illustrate and quantify this problem.
Here, the flue-damper system could cut the amount of infiltration air in half, at all firing rates. Assuming roughly equal time at high, medium and low fire throughout the day, the estimated fuel and O2 savings would be $47,000/year. Further, O2 and fuel savings could accrue by reducing the oxy/gas ratio, especially at low fire. Through these measures, reduced O2 concentration in the furnace atmosphere could also reduce dross formation (improve aluminum yield), providing additional dollar savings.
To summarize, flue gas measurements provided the following benefits:
- Establishes and quantifies the furnace atmosphere (oxidizing in this case)
- Identifies the excessive oxy/gas ratio at low fire
- Calculates infiltration air quantities at low, medium and high fire
- Aids in setting potentially improved flue-damper positions to reduce air infiltration
- Confirms that O2/gas ratios could be reduced
Well-Charged Aluminum Reverb Melter with Air/Oxy/Gas Burners
Flue-gas compositions were measured from a well-charged aluminum reverb melter firing with two air/oxy/gas burners. This customer was interested in determining the lambda setpoint at which the combustion atmosphere achieved neutral conditions. The decision was made to vary lambda over a wide range (lambda traverse) and compare measured flue-gas composition to calculated in a V curve similar to Figure 1 (in part 1).
This data is shown in Figure 3 for 13% O2 participation (87% air) at high fire (30.8 MMBTU/hour). Comparing actual flue gas (O2) and CO to calculated, the V curve is shifted to the right. As shown in Figure 3, a setpoint lambda of 1.05 was required to achieve a neutral combustion atmosphere.
Figure 3 indicates that, for whatever reason, this combustion system was getting about 5% less total oxygen (from air and O2 combined) than expected based on the combustion-system settings (setpoint lambda). This could be due to slight inaccuracy in the gas, O2 and/or combustion airflow meters, a slight inaccuracy in the fuel/oxidizer ratio controller, or the actual natural gas composition could be slightly different than the assumed 100% CH4 composition. From a purely practical standpoint, it does not matter what the exact reason is. In this particular case, the customer knew that they needed to set lambda at 1.05 to achieve a neutral combustion atmosphere (bottom of the V).
The flue damper was fully open, since these firing conditions (13% O2) have relatively high POC volumes, similar to air/gas (0% O2).
At these firing conditions (high fire, 13% O2), the furnace appeared to be at positive pressure, and since it was getting less total O2 (from combustion air and O2) than predicted based on combustion settings, it seemed reasonable to assume that there was no extra infiltration air. Indeed, calculated infiltration airflow rate for all data points shown in the Figure 3 V curve is zero (a calculated negative value for Ai is presumed to be zero).
The burners were subsequently converted to 90% O2 participation (10% air). Firing rate was reduced to 20.8 MMBTU/hour to account for the increased thermal efficiency (% available heat) with 90% O2. This increase in O2 participation results in lower burner POC (flue-gas volume). Not surprisingly, much higher flue gas (O2) was observed (12.5% vs. 1.1% O2), even with lambda reduced from 1.11 to 1.03. This suggests that infiltration air was present. Calculated infiltration air at these settings is about 40,000 SCFH.
The flue damper was then closed (in two steps). Flue gas (O2) level and calculated infiltration air were reduced with each step. Then lambda was reduced in order to reduce flue gas (O2) to 1%. This data is summarized in Table 4 and in Figure 5.
The test of this example represents an annual fuel and O2 savings of ~$138,000/year by adjusting the flue damper to reduce infiltration air by 30,500 SCFH. Again, the reduced O2 content in the furnace atmosphere would also be expected to provide reduced dross formation (improved aluminum yield).
In summary, for this aluminum reverb furnace example, flue-gas analysis was beneficially utilized to:
- Calibrate the combustion-control settings based on actual furnace atmosphere chemistry (V curve)
- Calculate infiltration-airflow rate in SCFH
- Help adjust the flue damper setting to reduce infiltration air
- Optimize combustion system settings to achieve desired furnace atmosphere conditions
For this furnace at low fire, similar to any other furnace, the flue damper should be closed more to minimize infiltration air at low fire.
Well-Charged Aluminum Reverb Fired with Waste Oil: Air/Oxy/Fuel Burners
Flue-gas composition was measured for a two-burner, well-charged aluminum reverb melt furnace, firing with recycled oil, with an air/oxy/oil combustion system. Oil heating value was 142,000 BTU/gal. The oil was electrically heated to about 140°F to provide the required “flowability.” The combustion-control system included oil flow meters (measuring oil flow in gpm) with closed-loop flow control of oil, O2 and combustion air. As is typical, fuel flow was controlled based on furnace temperature (fuel-lead) and the O2 and combustion airflow ratios followed at the desired flow-ratio settings.
In order to set the O2 and combustion airflows, the fuel flow was first converted to “natural gas equivalent” for the given MMBTU/hour input rate (measured oil gpm x heating value).
In this test, flue-gas analysis was utilized to make sure that proper O2 and airflow rates were determined and to aid in setting the flue damper position. The goal was to set lambda and the flue damper to maintain 3-4% flue-gas O2 at high fire. In this case, it was desirable to have a small excess of O2 to make sure that the oil was being completely combusted, recognizing there could be variation in oil composition or temperature, which could affect oil-flow measurement.
Some example flue-gas data are shown in Table 5.
For this furnace firing with waste oil, in order to maintain 3-4% O2 in the flue gas, the required lambda (excess oxidizer) setting was always higher than what one would calculate assuming equivalent natural gas composition. This indicates that this waste-oil fuel required a higher ratio of O2/BTU than natural gas and/or there may have been some absolute inaccuracy in the oil flow meter. (The exact waste oil composition was not known.)
This is especially significant considering that approximately 15,000 SCFH of compressed air was utilized for atomization, which adds to the combustion-air blower amounts shown.
This furnace appeared to be at positive pressure, and the draft gauge usually indicated slightly positive. Indeed, when calculating infiltration air, based on equations outlined in the Appendix (assumes CH4 fuel composition), negative or zero values were always determined. This suggests that there was no infiltration air and that all combustion air is supplied by blower (plus atomizing air). In this example for calculating infiltration air (A – Ac), it would be more accurate to utilize the measured CO2 and O2 values instead of the measured X and O2 value because the O2/fuel ratio X is less confidently known (liquid waste-oil fuel compared to gaseous O2 flow measurement).
In this example, the flue-gas composition measurements provided confirmation that the control system O2 and combustion-air ratio settings provided the desired combustion atmosphere, considering the unknown oil composition and any inaccuracies in flow measurement.
Glass Reverb Melt Furnace: 100% Oxy/Gas Burners with 91% Purity VSA O2
One final example is considered. Flue-gas compositions were measured from a glass reverb melt furnace, firing with 100% oxy/gas burners with VSA (on-site generated) oxygen of 91% purity. Firing rate is 14 MMBTU/hour (14,000 SCFH natural gas).
This operator was interested in determining the amount of infiltration air being pulled through the furnace and in fine-tuning the combustion system oxy/gas-ratio setting for minimum excess O2. This furnace was equipped with an automatic flue-damper control system (pressure control) to control furnace pressure to a desired setpoint. The purpose was to see if the furnace infiltration air could be further minimized.
In Figure 8, measured flue gas (O2) and (CO) is compared to calculated values, plotted versus oxy/gas ratio. To generate the graph, values for air infiltration (A = air/gas ratio) were assumed: A = 0-1.5, and calculated values for flue gas (O2) and (CO) were first plotted. Note, for A = 0 (zero infiltration air), an oxy/gas ratio of 2.20 is required for oxidizing conditions, owing to the 91% O2 purity. For the calculated values for A from 0-0.5, the V shaped curves show CO for reducing conditions and O2 for oxidizing conditions. Above A = 0.5, only oxidizing conditions are calculated (minimum oxy/gas ratio of 2.0).
During normal furnace operation, five data points for measured flue gas (O2) at oxy/gas ratios of 2.07, 2.10 and 2.15 are shown on the graph. All five data points fall very close to the calculated curve for A = 1.0. Therefore, it is reasonable to assume that this furnace was drawing infiltration air at a 1:1 ratio to the natural gas, or infiltration air = 14,000 SCFH.
Since this furnace was near the end of a refractory campaign, a cooling air blower was positioned near a bottom section of refractory where some cracks had developed. This blower was oriented so that the air blew directly into the furnace cracks. This blower was shut off for several minutes, and the flue gas analyzer readings were allowed to stabilize. With this air blower turned off, the furnace atmosphere conditions became reducing (zero [O2] with 2.8% [CO]). This was the only occasion when reducing conditions were measured.
This data point is also plotted on the Figure 8 graph. This data point ([CO] = 2.8%) falls on the calculated V curve (reducing section of the curve) for A = 0.35. So, we concluded that furnace infiltration air had been reduced from A = 1.0 to 0.35 (from 14,000-4,900 SCFH) by turning this blower off.
For practical purposes, this cooling blower could not be turned off permanently. However, it was re-positioned so that the air blew across the refractory at an angle instead of directly into the furnace.
Assuming that infiltration air can be cut in half (from Ai = 1 to 0.5) by re-positioning the blower, the calculated fuel savings would be about 4%, resulting in calculated natural gas and O2 savings of about $51,000/year.
This example demonstrates a more detailed graphical technique for analyzing flue-gas measurements and utilizing the data to calculate furnace air infiltration.
Summary of Calculated Infiltration Air Levels: All Air/Oxy/Fuel and Oxy/Fuel Furnaces
Comparing calculated infiltration air levels for all air/oxy/fuel and oxy/fuel furnaces considered, using the factor Ai (ratio of infiltration air to natural gas firing rate) to facilitate comparison, one can draw the following conclusions:
- At high fire, Ai is typically in the range from 0-1, and usually 0-0.5 after the flue damper has been adjusted. In two cases, Ai was in the range of 1-2 (reverb prior to damper adjustment, and oxy/gas rotary with lambda 0.9).
- At low fire, Ai is typically in the range of 1-3. One exception with much higher Ai was the oxy/gas reverb with very low turndown (9:1), and with misadjusted oxy/gas ratio. This is considered to be a unique case and not typical.
As illustrated in several examples from fuel-fired industrial melting furnaces, flue-gas analysis can be utilized to provide the following benefits:
- Calibrate the combustion-control system settings to provide the desired furnace atmosphere (oxidizing or reducing)
- Accurately quantify the extent of oxidizing or reducing conditions
- Diagnose combustion-system equipment problems or control-system mis-adjustments, offsets or inaccuracies
- Measure the quantity of furnace infiltration air. This is especially useful when comparing different firing rates and/or varying O2 participation levels.
- Aid in adjusting the flue damper for improved control of infiltration air (furnace pressure) and excess O2 level in the furnace atmosphere to provide energy cost savings and product quality and/or yield improvement.
Graphical techniques are presented which help to illustrate and quantify the combustion conditions (V curve) and provide insight when comparing measured flue gas composition to calculated (predicted) values.
These beneficial flue-gas analysis techniques are applicable to any fuel-fired heating or melting furnace, utilizing natural gas or other gaseous or liquid fuels, and employing air/fuel, oxy/fuel or air/oxy/fuel combustion systems.
Stewart Jepson, Air Liquide - Combustion Specialist, 800-820-2522