Experimental Investigation of Air–Fuel Mixing Effects on Flame Characteristics in a Direct fired Burner

Abstract

The length and pattern of air–fuel mixing plays a significant role in the uniformity, flame temperature, and emission characteristics, which can lead to a superior product quality in a non-oxidizing direct fired burner for a cold-rolled steel plate furnace. In this study, a diffusion-flame-type burner and partially-premixed-type burner were experimentally investigated to understand their effects on flame shape, flame temperature, and exhaust gas characteristics. With this aim, fuel nozzle size, nozzle hole number, fuel injection angle, and mixing distance of fuel and air were varied during the experiments. Computational fluid dynamics simulations were also performed to investigate the air–fuel mixing state for a nozzle-mixed burner and a partially-premixed burner. The results show that the flame temperature of the partially-premixed burner increases by up to 26 °C on average compared to that of the nozzle-mixed burner. It is also shown that the mixing distance plays an important role in the flame temperature of the partially-premixed burner. In addition, the residual oxygen concentration and volume ratio of CO/CO₂ in the flue gas of the partially-premixed burner exhibit lower concentrations compared to those of the diffusion flame burner.

1. Introduction

Recently, global attention and efforts have been focused on climate change and pollutant emission, and these efforts have sometimes led to strict regulations at the national level. One of the most affected industries in these regulations is the steel industry, which seeks to improve the efficiency of new manufacturing methods and energy in the process to address them. In this regard, these efforts are made to compact the facility within the manufacturing process. In the cold rolling process of steel, a heat treatment process is used to produce general cold-rolled and automobile steel sheets. This process consists of continuously suppressing the oxidation of the steel material surface and heating up the steel material to a required specific temperature for each product. The heating method in this heat treatment process is mainly divided into an indirect heating method, whereby a radiant tube is used, and a direct heating method, which uses a direct firing flame. The direct heating method can reduce the size of the furnace to about 1/5 of that of the indirect heating method, and hence can significantly reduce the energy required for heating the furnace. Consequently, this method allows the achievement of both fuel economy and facility compactness. The main function of the non-oxidizing direct fired furnace used in the heat treatment process of the cold-rolled steel plate is to prevent the excessive oxidation of the surface of steel and heat it up to 700–760 °C for steel moving at a constant line speed while maintaining the furnace temperature of 1100–1300 °C. Through this process, the material is preheated to the temperature for annealing. Simultaneously, impurities attached to the surface are eliminated and activated, enabling the oxide film reduction in the subsequent annealing process. The excessive oxidation of materials in the non-oxidizing direct fired furnace causes the oxide film to remain on the surface of the material after the annealing process, rendering it impossible to process surfaces, such as poor galvanizing, thereby resulting in a loss.

Unlike indirect heating methods using radiant tubes, the direct heating of the material is performed using a flame. Thus, depending on the combustion characteristics and conditions of the burner, the oxidation and reduction characteristics of the surface are directly affected. In particular, the temperature and composition distributions of combustion gases near the surface of the material are closely related to the oxidation and reduction behavior of the material, which is governed by the combustion characteristics of the burner.

Additionally, a reduction atmosphere should be maintained in the exhaust gas for the direct firing burner furnace to prevent oxidation of the steel material furnace near the fuel equivalence ratio (φ) of 1.0–1.25, which corresponds to a fuel-rich condition. Thus, product defects occur due to the localized oxidizing atmosphere not only by lack of precise control of the air–fuel quantity of the burner but also due to the aging of the burner. This eventually leads to an enormous economic loss. Park et al. [1] investigated the combustion characteristics and oxidation of the steel plate surface. They obtained a steep temperature gradient and the entrainment of residual oxygen. Premixed burner flames prevent the temperature drop of combustion gas on the steel plate surface. It was concluded that the elimination of residual oxygen and the temperature increase of combustion gas on the heating surface are essential to enhance the burner performance for application to non-oxidizing direct fired furnaces. Strommer et al. [2] investigated the combustion processes inside a direct fired continuous strip-annealing furnace. Here, the fuel-rich combustion natural gas was modeled by two chemical reactions: oxidation and water-gas-shift reactions, leading to a chemical equilibrium inside the furnace. Based on the combustion process, the flue gas dynamics were modeled by the mass and enthalpy balance. A measurement campaign conducted at a real plant was used for the model’s verification. The model was appropriate for real-time applications in control and optimization owing to its low computational effort and high accuracy. Banerjee et al. [3] investigated a methodology to control direct fired furnaces. It was reported that the control method required a model for heat transfer and combustion. Here, the application of the method showed that the fuel input and heat transfer to the load could be controlled to maintain a specified instantaneous temperature. This was accomplished through a feedback loop constructed by comparing the desired and measured temperatures. A parametric study was conducted to demonstrate the model.

Another important factor that influences flame stability is nozzle geometry. Elabaz et al. [4] performed the effect of the coflow on the stability and flame structure. They generated partially-premixed turbulent flames with a non-homogeneous jet of propane in a concentric flow conical nozzle burner. The results showed that the flame was stable between the two extinction limits of mixture inhomogeneity, and the optimum stability was obtained at a certain degree of mixture inhomogeneity. Milosavljevic et al. [5] established the flame characteristics of a swirl-stabilized, non-premixed natural gas burner as a function of the equivalence ratio and swirl number. Flame lengths decreased with increasing swirl and bulk air velocity, but the recirculation zone length was largely unaffected by the changes in these two quantities. Axial fuel injection through a centrally located nozzle resulted in symmetric flows only for fuel-lean flames such as swirl numbers below 0.6 and for near stoichiometric flames at swirl numbers above unity. The asymmetric flames were not due to asymmetric inlet velocity profiles. The angled annular or radial injection of natural gas extended the range of operation where symmetrical flames could be obtained, without greatly changing the flame lengths or stability characteristics. Zadghaffari et al. [6] investigated burner configuration to minimize pollutant emissions using numerical method. They presented computational fluid dynamics model of a non-premixed burner and focused on the study of the angle of gas, which was injected through the burner tip to the combustion zone. The present study finds the best geometry of the burner tip by testing the different angles of gas injection to optimize the burner performance and decreases the pollutant emission. The best geometry could bring significant benefits, such as low NOx and CO emissions and overall high combustion efficiency. Aggarwal [7] conducted a numerical and experimental study for the extinction and blowout of laminar partially-premixed flames motivated by the consideration of fire safety and suppression. Leung et al. [8] examined experimentally to obtain information on their thermal, emission, and heat transfer characteristics for two swirl-stabilized flames: a premixed flame and an inverse diffusion flame. Galley et al. [9] investigated a laboratory-scale swirling burner, presenting many similarities. Yang et al. [10] examined the influence of fuel preparation, combustor geometry, and operating conditions on combustion characteristics in the swirl-stabilized combustor. Wang et al. [11] investigated the potential role of a burner diameter in the flame structure and instability and the influence of burner diameter on the turbulent premixed flame structure.

Recently, studies have been conducted in various ways, not in the form of the existing type of burner. Lee et al. [12] performed the combustion characteristics of a porous media burner (PMB) with a large surface area for application in a non-oxidizing annealing furnace using a model furnace. The furnace temperature is controlled according to the variation of the thermal load and equivalence ratio (1.05–1.35), and its range was 1130–1330 °C at a 0% oxygen concentration or non-oxidizing atmosphere. In high furnace temperature conditions, the PMB surface temperature and radiant heat flux are also measured. Based on the results, the combustion mode changes, which implies a transition from the surface to the submerged combustion mode of the PMB that was directly compared with the crossover temperature. Finally, the PMB durability is evaluated through a cyclic on-off operation test and consequently, the operational safety was confirmed. Iman et al. [13] investigated NOx and thermal efficiency by assisting plasma to radiant tube burners. This design comprised several features of air staging, air swirling, and plasma-assisted combustion technology to improve performance and minimize NOx emissions. Not only in terms of modeling but also from a practical perspective, this study addresses the pressing need for the heat treatment industry to adopt air staging, air swirling, and plasma-assisted combustion in the radiant tube burner, which processes can ensure high heat recovery efficiency and low NOx emissions.

Premixed burners and nozzle-mixed burners have been generally used for the direct heating method. The premixed burner is more useful than the nozzle-mixed burner in terms of product quality despite having several drawbacks, such as backflash and a large number of burners due to their small size. Here, the excessive oxidation of material in a non-oxidizing furnace causes the oxide film to remain on the material’s surface after the annealing process, rendering it impossible to process surfaces, such as poor galvanizing, thereby resulting in production loss. Therefore, in this study, the effect of the nozzle-mixed burner and partially-premixed burner on air–fuel mixing, flame temperature, mixing with ambient gas, and exhaust gas were experimentally investigated.

2. Experimental Techniques and Facilities

2.1. Experimental Apparatus

The effect of the mixing distance on the flame was investigated by adjusting the distance between the injection nozzle tip and the point at which the air–fuel mixing process begins inside the burner. Figure 1 shows the schematic diagram of the experimental setup used in this study. The flame shape of the burner was recorded using a high-speed camera (Photron FASTCAM SA3 model, Japan), and the flame temperature was measured using an R-type thermocouple, moving it by 100, 200, and 300 mm from the burner exit.

A combustion gas analyzer (Testo 350K, Germany) was used to measure the concentration of O₂, CO, and CO₂ in the flame. A laser equipment (Optical system Co., Ltd., 532 nm, 2 W, Korea) and a smoke generator (Antari Co. Ltd., Taiwan) were used to visualize and investigate the homogeneous mixing degree of the air and fuel mixture gas. In addition, a commercial Star-CCM code was used to compare the mixing characteristics of the burner with the simulation results. This code (processing speed of 3.33 GHz) is frequently used for thermal flow field analysis. Liquefied natural gas (LNG) was supplied as a fuel from the gas bomb (which was pressurized at 120 bar and depressurized at 1 bar at the bomb exit) to the burners through a gas pressure regulator (Parker, Co, USA) with the final pressure of 300 mm H₂O. Air for the burner was supplied from an air blower of 400 mm H₂O with a rated power of 1.75 kW. Fuel and air quantity were controlled through mass flow controllers (Dwyer Instruments, IN, USA).

2.2. Burner

In this study, a nozzle-mixed-type burner was used, as shown in Figure 2a. Fuel was injected through eight fuel ejection holes (with diameter of 9.5 mm) from the nozzle center, and combustion air was induced through a swirling passage with an angle of 45° in Figure 2c. Fuel was finally injected through a nozzle exit with an angle of 15°. The swirl strength S could be obtained as S = tan θ, where θ is the angle of the swirl flow passage. The swirl number of the nozzle-mixed burner used in this study was approximately 1. Figure 2b shows a partially-premixed-type burner, where fuel and combustion air are premixed inside the burner. The injection and swirl angles of the partially-premixed burner were the same as for the nozzle-mixing-type burner, with the tip nozzle having two geometries (diameter × quantity of 8.5 mm × 10 and 9.5 mm × 8). However, just before injection, there is a difference in the partially-premixed burner in which fuel and air are mixed in advance in front of 116 mm. The low calorific value of LNG was 42.6 MJ/Nm³, and the main component is presented in

Table 1. LNG (Liquefied Natural Gas) composition.

Element	Unit
N2 (Nitrogen)	% mol.	0.12
CH4 (Methane)	% mol.	93.77
C2H6 (Ethane)	% mol.	3.80
C3H8 (Propane)	% mol.	1.67
I-C4H10 (Iso-Butane)	% mol.	0.29
N-C4H10 (Normal-Butane)	% mol.	0.32
I-C5H12 (Iso-Pentane)	% mol.	0.02

.

2.3. Experimental Techniques

The conditions used for the experiments and simulations are shown in

Table 2. Experimental conditions.

Mixing Type	Nozzle Type Diameter × Quantity	Fuel Equivalence Ratio, φ	LNG [Nm3/h]	Mixing Distance [mm]
Nozzle mixing	“a” 9.5 × 8	1.25	6	-
		1.1
		1.0
Partially premixing	“b” 9.5 × 8 “c” 8.5 × 10	1.25		80, 160, 240
		1.1
		1.0

. In each case, experiments and simulations were performed using the mixing type, fuel equivalence ratio, and mixing distance as variables. In nozzle mixing type, fuel flow rates for the burner were fixed at 6 Nm³/h and that of combustion air was varied to correspond to a fuel equivalence ratio of 1.0 to 1.25. In a partially-premixed type, fuel flow rates for the burner were fixed to 6 Nm³/h and that of combustion air was changed as a fuel equivalence ratio. The nozzle type was conducted in two types, i.e., “b” and “c”, to investigate the effect of the nozzle-mixed burner and partially-premixed burner on air–fuel mixing, flame temperature, mixing with ambient gas, and exhaust gas. R-type thermocouples were used to measure the flame temperature moving it by 100, 200, and 300 mm from the burner exit. A high-speed camera was used to take the photos of some flames.

3. Results and Discussion

Prior to the combustion experiment, the mixing characteristics of the air and fuel were investigated through photos taken using the smoke generator, the laser, and the high-speed camera, able to capture 500 frames per second. From Figure 3 it can be seen that the mixing effect of the nozzle-mixed-type burner (Figure 3a) is better than that of the partially-premixed-type burner (Figure 3b) when the mixing distance is 80 mm. This is also shown by the contrast image of the air–fuel mixture, which was processed as a black and white image.

The photos and images above suggest that, even for the partially-premixed burner, the air and fuel mix poorly when the mixing length is not large enough, which consequently leads to a heterogeneous temperature distribution in the flame. Here, by comparing the approximate area of the images in Figure 3a,b, we found that Figure 3a was 126% wider than Figure 3b, which confirms that Figure 3a is better mixed. For a quantitative review of Figure 3, the RGB (Red–Green–Blue) was compared on each of the same line. The comparison method was determined to be uniform with relatively few “B values”. It can be seen that Figure 3a is relatively 148% more homogeneous than Figure 3b. On the other hand, when the mixing length is increased to 160 and 240 mm, as shown in Figure 4, the degree of homogeneity of the air and fuel mixture increases, since air and fuel are well premixed at the nozzle front. In particular, the entire mixing degree of air and fuel exhibits superior results when compared to that of the conventional nozzle-mixed burner. Zhao et al. [14] investigated the effect of mixing distance on the mixing process and combustion of the double-cone burner. They obtained that the dimensionless mixing distance had a significant impact on the fuel mixing effect and then affected the high-temperature zone position and emissions using numerical methods. The natural gas–air mixing in the mixing chamber showed that the uniformity of methane concentration increases with an increase in distance from the ejector outlet.

Figure 5 shows photos and images for different air and fuel mixing lengths for the case with 10 nozzle holes of c-type, with a nozzle diameter of 8.5 mm. A less homogeneous air–fuel mixture exists in the vicinity of the nozzle exit regardless of the mixing length, whereas air and fuel are well mixed after a certain distance from the nozzle exit. This may be due to the fact that the air–fuel mixture is injected and then mixed more homogeneously with the surrounding gas via the improved entrainment effect through the increased nozzle holes number compared with the conventional nozzle under the same air–fuel velocity. However, in this case the flame temperature can be lowered due to flue gas recirculation into the flame since most of the surrounding is burned gas.

Figure 6 shows the maximum temperature of the cross section of each flame at different distances from the burner nozzle exit, obtained using the Star-CCM code for an air–fuel mixing distance of the three nozzles of 240 mm. The calculation area was limited to burners, and the combustion model used the eddy breakup model. The eddy breakup model is commonly utilized for pre-mixing, partially pre-mixing, and diffusion combustion, and selects a hybrid scheme that can simultaneously reflect turbulent mixing and chemical reaction rates. Three reaction control methods are provided for the EBU model, of which a hybrid is used when the combustion rate is controlled by both turbulent mixing and chemical reaction rate. The hybrid is applied because we consider changes in combustion chemical reactions with flow. Mesh applied polyhedral mesh. Star-CCM is a commercial tool built on FVM (Finite Volume Method) that supports polyhedral grid and helps to produce stable solutions when using the polyhedral grid. In this simulation, we applied a steady condition, segregated method. In Star-CCM, a simple algorithm is applied as the solver when using the steady and segregated method. Then, for the mesh convergence test, the number of meshes in the calculation domain was changed and checked by trial and error. In particular, the density of mesh in the center of the burner outlet and calculation area was changed, and the solution was checked for changes in density. Furthermore, the combustion parameters of the combustion model such as activation energy and temperature exponent were determined by trial and error through comparison with the experimental results. “a” refers to a nozzle-mixed-type burner (with eight nozzle holes), “b” denotes a partially-premixed-type burner (with eight nozzle holes), and “c” refers to a partially-premixed-type burner (with 10 nozzle holes). As can be seen from the temperature contour images in Figure 6, the flame shape of the partially-premixed-type burner is more rounded than that of the nozzle-mixed-type burner. In particular, the flame shape of the partially-premixed-type burner with 10 nozzle holes is almost a circle. This may originate from the fact that the air–fuel mixing process occurs more intensely than what described above.

A uniform heat transfer from the flame to the cold-rolled steel plate, which is a product, is important to reduce defects. Therefore, the flame temperature was here measured in open flame conditions for six selected measurement positions at the z = 300 mm point under a fuel equivalence ratio of φ = 1.1, which is mostly used in the real work process. As shown in Figure 7, the flame temperature is higher at positions #1, #2, and #3 than at positions #4, #5, and #6 regardless of the burner type, due to flame buoyancy in the atmosphere. In addition, it was observed that the average flame temperature of the partially-premixed-type burner “b” is 26 °C higher than that of the nozzle-mixed-type burner. This may be due to the fact that air and fuel are well premixed in the burner, which consequently leads to a rigorous combustion reaction. By contrast, in the case of the partially-premixed-type burner “c”, it was observed that the flame temperature is lower than that of the nozzle-mixed-type burner “a”. This may originate from the fact that the combustion reaction is retarded due to the rapid mixing of the air–fuel mixture with the atmospheric gas via the burned gas recirculation into the flame.

Generally, the fuel equivalence ratio in the non-oxidizing direct fired furnace of the cold rolling heat treatment process is maintained in a range between 1.0 and 1.25, to render the atmosphere gas in the furnace reductive.

A brief theoretical investigation was conducted on the oxidation and reduction of steel. In indirect heating using radiant tubes, reducing gas consisting of hydrogen and nitrogen is artificially formed to prevent steel surface oxidation, whereas in direct heating, combustion-induced gas acts as an atmospheric gas.

In general, the presence of residual oxygen by burning fuel in excess air is called an oxidative atmosphere. However, combustion gases produced under air shortage conditions are called reduction atmospheres. Because H₂O and CO₂ present in combustion gases oxidize the steel surface, steel surface oxidation occurs even if oxygen is not present in combustion gases. The value of H₂/H₂O and CO/CO₂ distinguishes oxidative and reductive atmosphere. From a chemical equilibrium perspective, the steel surface is oxidized by H₂O and CO₂ under a high-temperature atmosphere and reduced by H₂ and CO, i.e.,

Fe + H₂O ↔ FeO + H₂

(1)

Fe + CO₂ ↔ FeO + CO

(2)

In addition to oxidation and reduction reactions on these steel surfaces, surface carbon is precipitated and dissipated simultaneously, as shown in Equations (3) and (4).

C + CO₂ ↔ CO

(3)

C + H₂O ↔ CO + H₂

(4)

The precipitation and dissipation reactions are insignificant compared to oxidation and reduction reactions. The amount of residual carbon decreases rapidly when the material reaches the exit because of the temperature increase. The responses in Equations (1) and (2) involve equilibrium reaction constants, which are expressed as the functions of temperature. Generally, the values of H₂/H₂O and CO/CO₂ along with the steel surface temperature represent the extent to which the oxidation and reduction occur. As the steel temperature increases, the value of CO/CO₂ that can be reverted decreases while the value of H₂/H₂O decreases.

Figure 8 shows the area of oxidation and reduction depending on the equilibrium composition of combustion gases and the steel strip temperature. To heat the steel without oxidizing it to approximately 760 °C, the value of H₂/H₂O must be greater than or equal to 4.0, and the value of CO/CO₂ must be greater than or equal to 1.5. However, combustion must be performed in a fuel-rich condition with a fuel equivalence ratio of 2 or more. In the non-oxidizing direct fire, the fuel is burned in the range of 1.05 to 1.25. Thus, from an equilibrium perspective, the atmospheric gas is oxidative.

Katsuki et al. [15] examined the possibility that indirect heating by radiant tubes in a cold-rolled process in the steel industry may be replaced with direct flame heating without the steel plate surface oxidation. The possibility of its practical application has been proved although successful operating conditions are sensitive to controlling factors. They obtained that when the remaining oxygen concentration in the laminar and turbulent flame of methane fuel is less than 0.1%, the oxide film thickness for the steel sheet can be kept being less than 10 mm if the CO/CO₂ value becomes less than 0.4. From this viewpoint, the non-oxidizing direct fired furnace is responsively heating the material with minimal oxidation under a weak oxidative atmosphere, and the behavior of combustion products on the surface of the material, along with the temperature distribution, has a major effect on the oxidation and reduction.

Shigeru et al. [16] improved the descalability of annealed SUS304 steel using the vertical type direct fired heating furnace. The rapid heating of SUS304 steel with impinging flame burners at 673 to 1173 K decreased the accumulation of Fe and Si oxides into scale. However, soaking at a temperature above 1273 K for 10 to 20 s accumulated Cr and Mn oxides into the scale. Schmitz et al. [17] developed a burner that produced a reducing atmosphere in the furnace with a thermal efficiency comparable to state-of-the-art recuperative burners. The concept combined direct fuel-rich firing and indirect heating with radiant tubes in a recuperative burner. Unlike the existing recuperative burners, the concept-burners were equipped with an open radiant tube (ORT), forming an annular gap between the burner and tube, where the off-gas was post-combusted by adding secondary air. The reaction heat was either transferred to the furnace by ORT radiation or recuperated to heat the primary and secondary combustion air. In the presented numerical study, the impact of the primary equivalence ratio, the furnace temperature, and total equivalence ratio on the post-combustion process in the annular gap was evaluated. The total equivalence ratio had the most significant influence on post-combustion. The results for the variation of the total equivalence ratio showed that a faster post-combustion could be reached using a lower total equivalence ratio.

From this point of view, an experiment simulator for the cold rolling operation (with dimensions of 1.5 × 1.5 × 1.5 m) was fabricated to investigate the effect of the nozzle mixing type on the combustion behavior and combustion gas analysis. This investigation was performed for the nozzle-mixed-type burner “a” and partially-premixed-type burner “b”, which had a relatively high flame temperature. As can be seen in

Table 3. Comparison of O₂ concentration (%) and CO/CO₂ fraction.

Type	Fuel Equivalence Ratio (φ)	O2 (%)	CO/CO2
Nozzle mixing	1.25	0.12	0.36
	1.1	0.10	0.28
	1.0	0.01	0.25
Partially premixing	1.25	0.05	0.31
	1.1	0.02	0.25
	1.0	0.01	0.23

, satisfactory results were obtained for the two burner types using the criteria derived from the previous study. This suggests that problems should not be encountered in applying the partially-premixed-type burner to the actual process.

In addition, it was found that the partially-premixed burner tested in this study can be applied to the actual cold rolling operation, since both the O₂ concentration and that of CO/CO₂ via the fuel equivalence ratio are in good agreement with the combustion theory.

4. Conclusions

The air–fuel mixing process was experimentally investigated to understand its effect on flame temperature, mixing homogeneity, and exhaust gas concentration. The following conclusions could be drawn through experiments and simulations.

(1)

Through visualizations and simulations, it was shown that the mixing homogeneity and flame temperature increase when a specific mixing distance can be ensured for the partially-premixed-type burner “b.”

(2)

In this study, the partially-premixed-type burner “b” was relatively lower in terms of O₂ concentration and less than 0.4 in terms of CO/CO₂, indicating that this burner type can be applied to the real industrial field without significant formation of an oxide film.

(3)

Considering the characteristics of the cold-rolled process that must ensure a reduction atmosphere inside the furnace, it was found that the premixed-type burner is more favorable than the conventional nozzle-mix-type burner.