Professional Documents
Culture Documents
Bethany M. Worl1, Jianglin Fan1, Armin Silaen1, Jeffrey Cox2, Kurt Johnson3, Larry Fabina2, Joe Maiolo4, Kelly Tian4,
Chenn Q. Zhou1
1
Center for Innovation through Visualization and Simulation (CIVS), Purdue University Northwest
2200 169th Street, Hammond, IN 46323, USA
Phone: 1-219-989-2665
Email:czhou@pnw.edu
2
ArcelorMittal-USA
250 W US Hwy 12, Burns Harbor, IN 46304, USA
3
ArcelorMittal Global Research and Development
3001 E Columbus Dr, East Chicago, IN 46312, USA
4
Praxair, Inc.
175 E Park Dr, Tonawanda, NY 14150, USA
Keywords: Reheat furnace, computational fluid dynamics, CFD, oxygen enrichment, air/fuel ratio
INTRODUCTION
In 2017, the iron and steel industries in the U.S produced 23 million metric tons of pig iron and 82 million tons of steel; the
total value of iron and steel produced was $147 billion [1]. In the steelmaking process, the reheat furnace consumes the largest
amount of energy in the hot rolling stage and the second highest amount of energy in the entire process [2]. This energy
consumption is necessary for the heating of steel products up to the temperatures desired for plastic deformation in the hot
rolling stage of around 1150°C or greater [3]. The steel must also obtain a uniform temperature distribution to produce high
quality rolled steel and avoid damage to the rollers [4]. In order to reach these high temperatures and reduce the temperature
gradient, steel slabs may reside within a reheating furnace for over two hours. Reheated steel products can be in the form of
billets, blooms or slabs. There are many different avenues of optimization due to the myriad operational parameters found
within the reheating furnace.
The main objectives of reheating furnace optimization are to maintain satisfactory product quality, increase furnace
efficiency, increase productivity, and reduce emissions. The efficiency of the reheating furnace is commonly defined as the
amount of energy that enters the steel divided by the total furnace heat input. The efficiency can vary according to production
and delays, but is typically between 45-60%, with around 35-53% of the combustion energy released going to the slabs [5-8].
Product quality is quantified by the temperature distribution within the reheated slab but can also be adversely affected by
decarburization and oxidation within the furnace [9]. A temperature distribution within the steel can also damage the rollers
due to the differing plasticity within the slab. Within the steelmaking process, the reheat furnace is second to the blast furnace
in terms of greenhouse gas emissions, which include CO2, CO, and NOx, among other trace pollutants [5]. CO2 is a natural by-
product of the combustion process for hydrocarbons, whereas NOx is caused by excess oxygen reacting with nitrogen at the
very high temperatures found near the combustion flame. Production is also a value that many seek to optimize so long as
product quality does not suffer.
Improvement of thermal efficiency is often sought after as it should decrease emissions and increase productivity by nature of
its definition. The largest loss of heat from the furnace, and thus the largest detractor of the efficiency, is from the flue gases.
Most reheating furnace operations salvage energy from the flue gas using a recuperator to preheat the air being used in
combustion. Oxy-fuel and oxygen enriched combustion have also shown to be a somewhat novel method of reducing energy
loss by reducing the volume of the flue gas and the removal, in part or entirely, of nitrogen from the combustion reaction.
Reduction of nitrogen, typically 70% of the volume of outgoing flue gases, reduces possible NOx formation and leads to
Numerical models
Simulations were run using 3-dimensional, double precision, pressure based solvers for both the steady and transient cases
using ANSYS Fluent® CFD software. Mass and momentum conservation through their respective equations was ensured
through convergence criteria. The turbulence was modeled using the Reynolds Averaged Navier-Stokes (RANS) realizable k-
ε model. The energy equation was enabled to include heat transfer in the model. Necessary for combustion, the species
transport model with the eddy dissipation concept (EDC) model was used. The EDC model gives detailed chemical reactions
within the predicted flows. As radiation constitutes over 90% of the heat transfer into the slab, a discrete ordinates (DO)
radiation model was used in conjunction with the Weighted-Sum-of-Gray-Gases model (WSGGM) to capture the gray gas
behavior.
𝑥 𝑥 𝑡 (1)
where t is time, 𝑘 is the linear rate constant, and 𝑘 is the parabolic rate constant. The x term is either scale thickness or
weight gain per unit area, depending on the derivation of the rate constants and their subsequent units. The discretized form
of this equation is quickly found:
1
2
1
x ∑ ∙ ∆𝑡 (2)
1 2 4𝑡
𝑘𝑙 𝑘𝑝
where all terms retain their value from before, ∆𝑡 is the time step of the simulation, and t is the total time elapsed.
The heat transfer model for the scale formation made use of a modified thermal conductivity of the outermost layer of steel so
as to reflect the combined resistances of both the steel and the growing oxide layer. This was accomplished through an
effective thermal conductivity given by:
𝑘 (3)
where L is the thickness of either the steel or the scale (subscript s) and k is the thermal conductivity of the same. To be clear,
𝐿 is the thickness of the numerical cell being modified with the effective thermal conductivity by a user-defined function
(UDF) and is not the entire slab thickness. As this function takes the insulating effects of the iron oxide into effect, there is no
need to geometrically model the scale within the simulation.
Both of the scale formation model parts are implemented through the use of a UDF within ANSYS Fluent.
Primary fuel Secondary fuel Fuel Primary air Secondary air Air
mass flow rate mass flow rate temperature mass flow rate mass flow rate temperature
(ft3/s) (ft3/s) (°F) (ft3/s) (ft3/s) (°F)
Preheat zone
33.088 1.741 333.904 83.476
top
Preheat zone
26.514 1.395 271.792 67.948
bottom
Heat zone
13.936 0.733 131.424 32.856
top
80 811.3
Heat zone
9.756 0.513 92.096 23.024
bottom
Soak zone
2.251 0.118 23.112 5.778
small
Soak zone
4.370 0.230 41.232 10.308
large
Two different reaction mechanisms for the natural gas combustion were used in this simulation. The two-step Westbrook and
Dryer (WD) reaction mechanism was used for the study of air/fuel ratio, as it has been found to be an accurate assessment of
air-fuel flames [26]. Unfortunately, this accuracy is lacking when used to model oxygen enriched and oxy-fuel combustion.
Yin et al. found that the refined Jones and Lindstedt (JL) 4-step reaction model was better able to predict hydrogen levels,
flame temperature, and the carbon monoxide level in oxygen enriched combustion [27]. As such, the latter reaction mechanism
was used to investigate oxygen enrichment and oxy-fuel combustion undertaken in the latter half of the study. Both
mechanisms are summarized in Table 2 and in further detail in Ref. 27.
Westbrook and Dryer 2-step CH4 + 3/2 O2 → CO + 2 H2O 5.012 x 1011 2.0 x 108
mechanism CO + 1/2 O2 → CO2 2.239 x 1012 1.7 x 108
CH4 + 1/2 O2 → CO + 2 H2 4.4 x 1011 1.26 x 108
Refined Jones and Lindstedt CH4 + H2O → CO + 3H2 3.0 x 108 1.26 x 108
4-step mechanism H2 + 1/2 O2 ↔ H2O 5.69 x 1011 1.465 x 108
CO + H2O ↔ CO2 + H2 2.75 x 109 8.36 x 107
Scale formed once temperatures exceed 1292°F (700°C) has been found to be around 95% wüstite (FeO) [28]. As such, the
mathematical relationship for the thermal conductivity of wüstite above 550°C proposed by Akiyama et al. was used in the
calculation for the effective thermal conductivity [29].
4 22 2.56
The air/fuel ratio is an important factor with respect to combustion as it affects furnace gas temperature and resulting
emissions. Figure 2 details the change in the bulk temperature of the slab throughout the entire furnace residence time of each
case. In progressive order, the boundaries of the preheating, heating, and soak zones are denoted by the dashed vertical lines
at 76 minutes and 117 minutes. The model predicts a negligible increase in the bulk slab temperature profile with increasing
AFR until the slab approaches the heating zone. The largest temperature difference was found at the end of the furnace
residence time and was most likely due to changes in the gas temperature in the soak zone. Case 1 has the lowest slab bulk
temperature of 2287°F, which was expected for the fuel-rich substoichiometric combustion ratio. Fuel, entering the furnace
at ambient temperatures is heated to the furnace gas temperature but remains uncombusted and so does not provide heating
energy to the slab. Conversely, the AFR of case 2 that is closest to stoichiometric gives the highest slab bulk temperature of
2352°F. Increased air fraction results in lower gas temperatures as unreacted oxygen and higher nitrogen ballast dilute the
energy of combustion.
2500
Preheating Heating Soak
Slab Bulk Temperature (°F)
2000
AFR: 16
1500
2400 AFR: 18
2350 AFR: 20
1000
2300
AFR: 22
2250
500
2200
140 145 150 155
0
0 50 100 150 200
Residence Time (min)
Figure 2. Slab bulk temperature through furnace residence for AFR cases
0.08 2360
2350
0.07
Final scale thickness (in)
0.06 2330
2320
0.05 2310
20% Excess Air
10% Excess Air
0% Excess Air
(AFR = 17.6)
(AFR = 20.6)
(AFR = 18.9)
2300
0.04
2290
0.03 2280
14 16 18 20 22 24
Air/fuel ratio
Figure 3. Comparison of slab top temperature and final scale thickness with AFR
The combustion profile is also affected by a change in the AFR. Using CO concentration as an indicator of the extent of
combustion provides insight into the mixing of oxygen and fuel. Figure 4 shows the decreasing trend in the maximum CO
concentration near the burner with increasing AFR. This information was taken from a plane 2 inches from the burner face
where mixing between the fuel and air was underway. At larger AFR the extent of combustion increased as indicated by the
decrease in CO concentration. The lowest extent of combustion is at the highest CO concentration corresponding to the
substoichiometric AFR. The decrease in CO concentration between AFR 16 and 18 is at first steep but then begins to lessen
with each increment upwards of the AFR. Each upward increment of air/fuel ratio (mass basis) shown (i.e. 16 18) is
associated with a constant increase in the molar oxygen/fuel ratio of 0.23. Increased AFR also diluted this enhanced mixing
with nitrogen, which may explain the decrease in slope with increased AFR.
Maximum CO Concentration
2.30E-07
2.20E-07
2.10E-07
(kmol/ft3) 2.00E-07
1.90E-07
1.80E-07
1.70E-07
1.60E-07
1.50E-07
16 17 18 19 20 21 22
AFR
Figure 4. CO concentration variation near the preheat top zone burner
Flame shape and temperature was also investigated. Figure 5 shows the side profile of the preheat zone top flames for each
AFR case. The temperatures found at the track point follow the same trend as the slab surface and bulk temperatures, wherein
the lowest temperature was found in the substoichiometric case and the highest in the slightly superstoichiometric case of
AFR 18. Of note was the high temperature zone in excess of 3400°F found in the 22 AFR case; this flame was extremely
premature and could damage the burner and surrounding refractory. This premature combustion was likely caused due to the
large air/fuel ratio. The track point temperature is still higher than the temperature found in the substoichiometric case. The
temperature for case 2 is notably higher than that of case 3 with a 128°F difference; this is most likely due to the longer flame
shape found in case 2.
Figure 5. Flame profiles in the top preheat zone with track point shown
Oxygen Enrichment
Oxygen-enriched combustion (OEC) has various noted benefits [11, 31-32]. With OEC, the total amount of nitrogen in the
furnace is reduced; this has the potential to lower nitrogen oxide (NOx) emissions. Other emissions may also be reduced, such
as carbon monoxide and hydrocarbons. OEC is also purported to increase efficiency as the mass of gases leaving the furnace
is reduced and less energy is required to bring nitrogen up to the temperature of the combustion products; radiative heat
transfer is also enhanced. Productivity can also see an increase when fuel flow rate is held constant due to higher flame
temperatures and increased heat transfer to the slabs. The stability of the combustion itself has also been found to improve
due to better mixing of the oxygen and the natural gas.
Liu et al. conducted a small parametric study to see the effects of oxygen enrichment on the furnace gas temperature [24]. The
volume fraction of oxygen within the combustion air was increased and decreased by 2% for a total of 3 cases. They found
2600
2500
2400
Gas Temperature (°F)
2300
2200
2100
2000
18.9% oxygen
1900 20.9% oxygen
22.9% oxygen
1800
0 10 20 30 40 50 60 70 80 90 100 110
Distance from Charge Door (ft)
Figure 6. Gas temperature at the center face of the furnace with different oxygen volume fraction (Liu et al.)
In the present study, a more in-depth and properly implemented parametric study has been conducted according to Table 4 in
alignment with the presented AFR results above; here, the ratio of total oxygen to fuel was held constant while the level of
oxygen enrichment of the air changed. Specifically, this necessitated decreasing the flow rate of the enriched air to ensure
that the oxygen fuel ratio stayed constant at the same fuel rate. The ‘base’ case represents air-fuel combustion, and cases 1, 2
and, 3 are a low oxygen enrichment level, medium oxygen enrichment level, and oxy-fuel respectively. For the present study,
only case 1 is discussed in comparison with the base case. Higher oxygen enrichment levels will be discussed at a future
time.
Oxygen enrichment for the low oxygen enrichment case was applied to the preheat zone, so the investigation of slab bulk
temperature and any scale formation insights are based around the residence time of the slab from charging to the end of
preheating. Figure 6 shows the predicted slab heating profile of the low oxygen enrichment case to the base stoichiometric
case. The temperature profiles are similar with the base line case staying slightly hotter than the oxygen enriched case for the
first hour of residence time. Near the end of the zone, the slab bulk temperature of the low oxygen enrichment case (case 1)
begins to overtake the base case.
1400
1000
800
600
400
1.00E-02
6.00E-03
4.00E-03
2.00E-03 Base
Low Oxygen Enrichment
0.00E+00
0 10 20 30 40 50
Time (min)
Figure 7. Scale thickness between a low oxygen enriched combustion case and an air-fuel combustion case (base)
CONCLUSIONS
A 3D CFD model of an industrial reheating furnace has been developed and used to investigate the effect of varied air/fuel
ratio and oxygen enrichment on combustion performance. Furthermore, a previously developed scale formation model has
been applied to investigate the effects of changing the composition and amount of combustion air going into the furnace on
scale formation [21]. As expected, varying air/fuel ratio (AFR) caused changes in predicted slab bulk temperature and in scale
formation. Up to a limit, an increase in air rate into the furnace at higher AFR showed an increase in extent of combustion
and predicted an increase in scale thickness. Slightly superstoichiometric operation is where the highest slab temperature was
found. This was attributed to improved mixing that led to more complete combustion. Above an estimated AFR of 20, the
benefits of mixing are outweighed by the excess nitrogen in the ballast air and cause the furnace gas and the subsequent slab
temperature to decrease.
An examination of oxygen enrichment was modelled to further investigate the effects of combustion on reheat furnace
productivity and scale formation. Results show that slight oxygen enrichment increased the slab bulk temperature by 30°F
compared to the base case. As the amount of fuel flowing into the furnace was the same for each case, a slight improvement
in combustion efficiency (in terms of the ratio of heat absorbed into the slab over the total furnace heat input) was noted.
Comparing scale formation as a function of residence time that the average slab surface temperature was above 1292°F
(700°C), the model predicted a decrease in the rate of scale formation in the case of oxygen enrichment compared to the base
case. The base case had larger scale growth due to the longer residence time above the oxidation temperature limit and the
higher gas velocity near the slab. Further oxygen enrichment studies are planned in order to investigate the impact of higher
levels of oxygen enrichment.
ACKNOWLEDGEMENTS
We would like to thank the Steel Manufacturing Simulation & Visualization Consortium (SMSVC) members for funding this
project as well as their technical feedback. We would also like to thank the members of the Center for Innovation through
Visualization and Simulation (CIVS) for their knowledge and input throughout the project.
REFERENCES
1. “Mineral commodity summaries 2018,” U.S. Geological Survey, 2018, p. 82.
2. S.H. Han, S.W. Baek, M.Y. Kim, “Transient radiative heating characteristics of slab in a walking beam type reheating
furnace,” International Journal of Heat and Mass Transfer, Vol. 52, No. 3-4, January 2009, pp. 1005-1011.
3. P.V. Barr, “The development, verification, and application of a steady-state thermal model for the pusher-type reheat
furnace,” Metallurgical and Materials Transactions B, Vol. 26, No. 4, August 1995, pp. 1543-1916.