MILD combustion has emerged as a promising technology for achieving uniform temperature distribution, reduced pollutant formation, and improved thermal efficiency in industrial reheating furnaces. However, the performance of different burner geometries under MILD conditions—particularly in full-scale steel‑heating scenarios—has not been adequately characterized. In this study, the thermal behavior of three burner configurations, namely a co‑flow axial burner, a swirl burner, and a perforated lining burner, was experimentally investigated in a laboratory-scale MILD furnace developed at Tarbiat Modares University. The primary objective was to

Steel reheating furnaces are critical in metallurgical industries for preparing billets or slabs prior to hot forming. Conventional combustion systems in these furnaces often create large temperature gradients and emit high levels of nitrogen oxides (NOₓ). Flameless oxidation, also known as Moderate or Intense Low‑oxygen Dilution (MILD) combustion, has emerged as a promising alternative. By strongly diluting fuel and oxidizer with recirculated combustion products before ignition, MILD combustion suppresses visible flame formation, reduces peak temperatures, and promotes a more uniform thermal field within the furnace volume.
Despite these advantages, the performance of MILD combustion systems is highly sensitive to burner geometry and the resulting mixing characteristics of fuel and oxidizer streams. Different burner configurations can produce distinct flow structures, internal recirculation rates, and temperature distributions. Therefore, systematic experimental evaluation of burner designs is essential for optimizing industrial reheating furnaces operating under MILD conditions.
The present work addresses this need by experimentally comparing three burner types—co‑flow axial, swirl, and perforated lining—in a purpose‑built laboratory‑scale MILD furnace. The study focuses on both heating rate and flame stability, providing practical insights for burner selection in steel reheating applications.
Methodology
Laboratory‑Scale MILD Furnace
Experiments were conducted in a laboratory‑scale flameless oxidation furnace designed and constructed at the National Combustion Laboratory of Tarbiat Modares University. The furnace was scaled down from a 7 MW industrial steel reheating furnace using the constant residence time (CRT) scaling method, resulting in a 10 kW laboratory model. Key dimensions of the scaled furnace are 919 mm (length) × 513 mm (width) × 293 mm (height). The furnace walls and roof were insulated with multi‑layer refractory materials (cordierite‑mullite, lightweight firebrick, and perlite) to maintain external surface temperatures below 80 °C during operation up to 1600 °C.
Burner Configurations
Three distinct burner geometries were fabricated and tested:
· Co‑flow axial burner: A 1‑inch diameter burner with an axially adjustable fuel tube. The fuel nozzle could be positioned upstream or downstream of the air exit to control premixing and flame stability. Optimal stability was achieved with the fuel tube retracted by 2.5 cm relative to the air outlet.
· Swirl burner: A 1‑inch burner featuring a helical swirler (4 cm pitch) along the fuel tube. Air flowing over this spiral acquires a tangential velocity component, inducing a swirling motion at approximately 45° relative to the burner axis. This enhances mixing, flame volume, and stability.
· Perforated lining burner: A 1‑inch burner consisting of a perforated inner tube (½‑inch, 16 holes of 3 mm diameter) enclosed within a perforated outer casing (1‑inch, holes of ~1.5 mm, porosity ≈ 0.26). The design promotes distributed, multi‑point injection for uniform heating and increased power capacity (up to 30 kW).
Instrumentation and Procedure
Each burner was installed separately in the furnace, and tests were performed under comparable firing rates and air preheating conditions. Fuel (natural gas) and air flow rates were controlled via calibrated rotameters and pressure regulators. Temperatures were measured using S‑type thermocouples (for high‑temperature regions up to 1600 °C) and J‑type thermocouples (for stack and lower‑temperature zones). Thermocouples were arranged in a 9‑column, 3‑row grid on the sidewall to capture spatial temperature distributions. A CCD camera recorded flame behavior, and RGB image analysis was employed to assess flame uniformity and thermal structure.
Results and Discussion
Flame Structure and Stability
The co‑flow axial burner produced a long, narrow flame (≈35 cm) with relatively low momentum. The flame exhibited sensitivity to hot gas recirculation from the furnace rear wall, leading to intermittent instabilities. Due to insufficient mixing at an equivalence ratio of ~1.1, unburned gases were observed in the furnace.
In contrast, the swirl burner generated a voluminous, helical flame (≈40 cm) with significantly enhanced mixing and momentum. The swirling motion stabilized the flame, prevented lift‑off, and improved heat transfer to the furnace walls. However, increased turbulence led to moderate flame oscillations over time.
The perforated lining burner, when operated at 18.4 kW, produced a uniform, continuous flame sheet across its surface. During initial startup, distinct flame plumes were visible, but as the burner body heated up, the plumes diminished, and combustion transitioned toward a distributed, nearly flameless regime—characteristic of high‑temperature air combustion (HiCOT). Unfortunately, prolonged exposure to high temperatures caused severe surface oxidation and eventual localized fracture of the burner tube, as shown in Figure 12 of the full manuscript.
Thermal Performance and Heating Rate
Figure 14 of the original study illustrates the temporal response of furnace roof temperature for each burner:
· The co‑flow axial burner (10 kW) required approximately 450 minutes (7.5 hours) to reach 500 °C and exhibited no further temperature increase, indicating limited thermal capacity.
· The perforated lining burner (18.4 kW) rapidly heated the furnace, reaching 550 °C in just 150 minutes (2 hours 33 minutes). However, the test was terminated prematurely due to burner structural failure.
· The swirl burner (10 kW) achieved 550 °C after approximately 7 hours and surpassed 600 °C after 10 hours, with a continuing upward trend. This demonstrates reliable and sustained heating capability.
Stack gas temperature measurements (Figure 15) corroborated these trends: the swirl burner achieved the highest exhaust temperature (~213 °C) with an ongoing rising trend, whereas the co‑flow burner plateaued at lower values.
RGB Image Analysis
RGB intensity scatter plots (Figure 13) provided quantitative insight into flame uniformity. The co‑flow burner displayed the most concentrated and consistent color distribution, suggesting a homogeneous thermal field at the captured moment, though long‑term stability remained problematic. The swirl burner showed increased scatter—particularly in the blue channel—indicating the presence of turbulent, multi‑temperature zones. The perforated lining burner exhibited the widest dispersion, reflecting significant local temperature fluctuations inherent to its multi‑jet design.
Conclusions
This experimental investigation compared the performance of co‑flow axial, swirl, and perforated lining burners in a laboratory‑scale MILD furnace for steel reheating. The following conclusions are drawn:
· Co‑flow axial burner: Produces an elongated flame but suffers from low momentum and instability when exposed to recirculating hot gases. Its heating capacity is insufficient for reaching target temperatures within a reasonable timeframe.
· Perforated lining burner: Offers the fastest heating rate, achieving 550 °C in 2 hours 33 minutes. However, severe oxidation and structural failure at elevated temperatures render it unsuitable for long‑term industrial use without expensive alloy upgrades.
· Swirl burner: Provides the most stable and uniform heating performance. Although its heating rate is slower than that of the linear burner, it reliably attains and exceeds 600 °C without structural degradation. Scaling this design to larger diameters (e.g., 1.5–2 inches) could further improve heating rate while maintaining stability.
Overall, the swirl burner emerges as the most practical and robust option for industrial MILD furnaces, balancing thermal uniformity, operational safety, and long‑term durability. The findings underscore the critical influence of burner geometry on furnace startup time and temperature distribution, offering valuable guidance for optimizing burner selection in steel reheating applications.
Author Contributions
Ali Ashouri: Conceptualization, Methodology, Investigation, Formal analysis, Writing – original draft, Writing – review & editing, Visualization, Project administration.
Soroush Sarrafan Sadeghi: Investigation, Formal analysis, Data curation, Writing – review & editing, Visualization.
Mohammad Zabetian Targhi: Supervision, Conceptualization, Resources, Writing – review & editing, Funding acquisition.
Funding
This research received no specific grant from any funding agency in the public, commercial, or not‑for‑profit sectors. The study was conducted entirely at Tarbiat Modares University, Tehran, Iran, using the facilities of the National Combustion Laboratory.
Conflict of Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.


Steel reheating furnaces are critical in metallurgical industries for preparing billets or slabs prior to hot forming. Conventional combustion systems in these furnaces often create large temperature gradients and emit high levels of nitrogen oxides (NOₓ). Flameless oxidation, also known as Moderate or Intense Low‑oxygen Dilution (MILD) combustion, has emerged as a promising alternative. By strongly diluting fuel and oxidizer with recirculated combustion products before ignition, MILD combustion suppresses visible flame formation, reduces peak temperatures, and promotes a more uniform thermal field within the furnace volume.
Despite these advantages, the performance of MILD combustion systems is highly sensitive to burner geometry and the resulting mixing characteristics of fuel and oxidizer streams. Different burner configurations can produce distinct flow structures, internal recirculation rates, and temperature distributions. Therefore, systematic experimental evaluation of burner designs is essential for optimizing industrial reheating furnaces operating under MILD conditions.
The present work addresses this need by experimentally comparing three burner types—co‑flow axial, swirl, and perforated lining—in a purpose‑built laboratory‑scale MILD furnace. The study focuses on both heating rate and flame stability, providing practical insights for burner selection in steel reheating applications.
Methodology
Laboratory‑Scale MILD Furnace
Experiments were conducted in a laboratory‑scale flameless oxidation furnace designed and constructed at the National Combustion Laboratory of Tarbiat Modares University. The furnace was scaled down from a 7 MW industrial steel reheating furnace using the constant residence time (CRT) scaling method, resulting in a 10 kW laboratory model. Key dimensions of the scaled furnace are 919 mm (length) × 513 mm (width) × 293 mm (height). The furnace walls and roof were insulated with multi‑layer refractory materials (cordierite‑mullite, lightweight firebrick, and perlite) to maintain external surface temperatures below 80 °C during operation up to 1600 °C.
Burner Configurations
Three distinct burner geometries were fabricated and tested:
· Co‑flow axial burner: A 1‑inch diameter burner with an axially adjustable fuel tube. The fuel nozzle could be positioned upstream or downstream of the air exit to control premixing and flame stability. Optimal stability was achieved with the fuel tube retracted by 2.5 cm relative to the air outlet.
· Swirl burner: A 1‑inch burner featuring a helical swirler (4 cm pitch) along the fuel tube. Air flowing over this spiral acquires a tangential velocity component, inducing a swirling motion at approximately 45° relative to the burner axis. This enhances mixing, flame volume, and stability.
· Perforated lining burner: A 1‑inch burner consisting of a perforated inner tube (½‑inch, 16 holes of 3 mm diameter) enclosed within a perforated outer casing (1‑inch, holes of ~1.5 mm, porosity ≈ 0.26). The design promotes distributed, multi‑point injection for uniform heating and increased power capacity (up to 30 kW).
Instrumentation and Procedure
Each burner was installed separately in the furnace, and tests were performed under comparable firing rates and air preheating conditions. Fuel (natural gas) and air flow rates were controlled via calibrated rotameters and pressure regulators. Temperatures were measured using S‑type thermocouples (for high‑temperature regions up to 1600 °C) and J‑type thermocouples (for stack and lower‑temperature zones). Thermocouples were arranged in a 9‑column, 3‑row grid on the sidewall to capture spatial temperature distributions. A CCD camera recorded flame behavior, and RGB image analysis was employed to assess flame uniformity and thermal structure.
Results and Discussion
Flame Structure and Stability
The co‑flow axial burner produced a long, narrow flame (≈35 cm) with relatively low momentum. The flame exhibited sensitivity to hot gas recirculation from the furnace rear wall, leading to intermittent instabilities. Due to insufficient mixing at an equivalence ratio of ~1.1, unburned gases were observed in the furnace.
In contrast, the swirl burner generated a voluminous, helical flame (≈40 cm) with significantly enhanced mixing and momentum. The swirling motion stabilized the flame, prevented lift‑off, and improved heat transfer to the furnace walls. However, increased turbulence led to moderate flame oscillations over time.
The perforated lining burner, when operated at 18.4 kW, produced a uniform, continuous flame sheet across its surface. During initial startup, distinct flame plumes were visible, but as the burner body heated up, the plumes diminished, and combustion transitioned toward a distributed, nearly flameless regime—characteristic of high‑temperature air combustion (HiCOT). Unfortunately, prolonged exposure to high temperatures caused severe surface oxidation and eventual localized fracture of the burner tube, as shown in Figure 12 of the full manuscript.
Thermal Performance and Heating Rate
Figure 14 of the original study illustrates the temporal response of furnace roof temperature for each burner:
· The co‑flow axial burner (10 kW) required approximately 450 minutes (7.5 hours) to reach 500 °C and exhibited no further temperature increase, indicating limited thermal capacity.
· The perforated lining burner (18.4 kW) rapidly heated the furnace, reaching 550 °C in just 150 minutes (2 hours 33 minutes). However, the test was terminated prematurely due to burner structural failure.
· The swirl burner (10 kW) achieved 550 °C after approximately 7 hours and surpassed 600 °C after 10 hours, with a continuing upward trend. This demonstrates reliable and sustained heating capability.
Stack gas temperature measurements (Figure 15) corroborated these trends: the swirl burner achieved the highest exhaust temperature (~213 °C) with an ongoing rising trend, whereas the co‑flow burner plateaued at lower values.
RGB Image Analysis
RGB intensity scatter plots (Figure 13) provided quantitative insight into flame uniformity. The co‑flow burner displayed the most concentrated and consistent color distribution, suggesting a homogeneous thermal field at the captured moment, though long‑term stability remained problematic. The swirl burner showed increased scatter—particularly in the blue channel—indicating the presence of turbulent, multi‑temperature zones. The perforated lining burner exhibited the widest dispersion, reflecting significant local temperature fluctuations inherent to its multi‑jet design.
Conclusions
This experimental investigation compared the performance of co‑flow axial, swirl, and perforated lining burners in a laboratory‑scale MILD furnace for steel reheating. The following conclusions are drawn:
· Co‑flow axial burner: Produces an elongated flame but suffers from low momentum and instability when exposed to recirculating hot gases. Its heating capacity is insufficient for reaching target temperatures within a reasonable timeframe.
· Perforated lining burner: Offers the fastest heating rate, achieving 550 °C in 2 hours 33 minutes. However, severe oxidation and structural failure at elevated temperatures render it unsuitable for long‑term industrial use without expensive alloy upgrades.
· Swirl burner: Provides the most stable and uniform heating performance. Although its heating rate is slower than that of the linear burner, it reliably attains and exceeds 600 °C without structural degradation. Scaling this design to larger diameters (e.g., 1.5–2 inches) could further improve heating rate while maintaining stability.
Overall, the swirl burner emerges as the most practical and robust option for industrial MILD furnaces, balancing thermal uniformity, operational safety, and long‑term durability. The findings underscore the critical influence of burner geometry on furnace startup time and temperature distribution, offering valuable guidance for optimizing burner selection in steel reheating applications.
Author Contributions
Ali Ashouri: Conceptualization, Methodology, Investigation, Formal analysis, Writing – original draft, Writing – review & editing, Visualization, Project administration.
Soroush Sarrafan Sadeghi: Investigation, Formal analysis, Data curation, Writing – review & editing, Visualization.
Mohammad Zabetian Targhi: Supervision, Conceptualization, Resources, Writing – review & editing, Funding acquisition.
Funding
This research received no specific grant from any funding agency in the public, commercial, or not‑for‑profit sectors. The study was conducted entirely at Tarbiat Modares University, Tehran, Iran, using the facilities of the National Combustion Laboratory.
Conflict of Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.