- Email: [email protected]
Development of an oxygen-enhanced combustor for scrap preheating in an electric arc furnace
Applied Thermal Engineering 91 (2015) 749e758
Contents lists available at ScienceDirect
Applied Thermal Engineering journal homepage: www.elsevier.com/locate/apthermeng
Research paper
Development of an oxygen-enhanced combustor for scrap preheating in an electric arc furnace Jeongseog Oh*, Eungyeong Lee, Dongsoon Noh Advanced Combustion Laboratory, Korea Institute of Energy Research, Daejeon, Republic of Korea
a r t i c l e i n f o
a b s t r a c t
Article history: Received 25 June 2015 Accepted 30 August 2015 Available online 5 September 2015
The performance of a pilot-scale oxygen-enhanced furnace for scrap preheating with a multi-nozzle inverse diffusion combustor was studied. Natural gas and oxygen diluted in air were supplied to the oxygen-enhanced combustor with a range of 120e300 Mcal/h. The internal flow field of the furnace was simulated with a computational fluid dynamics program before any experiments were carried out. Meanwhile, the characteristic time for furnace heat-up and the level of pollutant emission were experimentally measured. Results showed that the temperature rise in the furnace increased steeply with an increase in the oxygen mole fraction in an oxidant. In addition, the characteristic time for furnace heat-up decreased exponentially as the thermal input power increased. © 2015 Elsevier Ltd. All rights reserved.
Keywords: Oxy-fuel combustion Oxygen-enhancement CH4-air flame Scrap preheating Electric arc furnace
1. Introduction Air-used combustion is a method that is commonly used in the steel melting or reheating processes. In recent, oxygen-enhanced or oxygen-enriched combustion (OEC) is introduced to increase thermal efficiency [1]. Steel and iron have served as basic materials for industry. The steel making process is classified as either blast furnace or electric arc furnace (EAF) method. Compared to the blast furnace, on advantage of the EAF method is increased energy savings due to the simple process and the low emission of carbon dioxide (CO2) [2]. In particular, this is relevant today because of the world wide attempts to reduce CO2 emissions due to global warming and climate change [3]. A new method for an effective thermochemical steel-making using in EAF process is needed to reduce energy consumption. Several alternatives that have been suggested use heat recovery from burned gas (i.e. flue gas), such as the Eco-Arc and Consteel processes [4,5]. The core concept of both methods is scrap preheating up to 600 C with high temperature flue gas (about 1200 C). As a secondary thermal energy source into relatively cold zone, OEC and carbon lancing play an important role in increasing electric energy savings. This thermochemical method is one of
* Corresponding author. Advanced Combustion Laboratory, Korea Institute of Energy Research, Daejeon 305-343, Republic of Korea. Tel.: þ82 42 860 3479; fax: þ82 42 860 3133. E-mail address: [email protected] (J. Oh). http://dx.doi.org/10.1016/j.applthermaleng.2015.08.088 1359-4311/© 2015 Elsevier Ltd. All rights reserved.
ways to reduce electricity usage by removing the energy conversion process of fossil fuel energy to electric energy. OEC is characterized by a higher flame temperature than that of air-used combustion because a higher oxygen concentration in an oxidant leads to a lower absorption energy by nitrogen (N2). This characteristic of the high temperature has been used in heat treatment furnaces, waste incinerators, propulsion engines, etc. Compared to air-used combustion, OEC is known to increase reaction rate due to the higher flame temperature and oxygen mole fraction in the same flow condition. Wu et al. [6] reported that a small amount of oxygen added to an oxidant (e.g. 21%e30%) resulted in a fuel consumption decrease of 26.1%. Increased OEC is good for not only energy savings, but also for productivity improvement. Bĕlohradský et al. [7] reported that combustion efficiency is influenced by the mole fraction of O2 in an oxidant ðXO2 Þ due to an increased CO2 and water vapor (H2O) mole fraction in flue gas. A flue gas temperature is in inverse proportion to the time needed for the heat transfer to the object being heated, such as iron nuggets and ingots. However, a weakness of OEC is the additional costs related to the increased oxygen supply and burner retrofitting because the increased combustion intensity leads to a high temperature environment in the furnace and the exposure of the burner surface to this heat level results in a need for more frequent maintenance. A method using CO2 recirculation has been suggested to control reaction rate and flame temperature [8]. CO2 dilution to an oxidant is effective in reducing the production of thermal nitrogen oxides (NOx).
750
J. Oh et al. / Applied Thermal Engineering 91 (2015) 749e758
Nomenclature a.u. CCS cpi DO dA dF dOx EAF EDM EDC EIi LHVi lpm MFC MWi mi NG NTP OEC QTh P∞ DP RANS
arbitrary unit carbon capture and storage specific heat of species i (cal/(g K)) discrete ordinate model outlet diameter in an air nozzle exit (mm) outlet diameter in a fuel jet nozzle exit (mm) outlet diameter in an oxygen nozzle exit (mm) electric arc furnace eddy-dissipation model eddy-dissipation concept model emission index of species i (g/kg) low heating value of species i (kcal/Nm3) liter per minute mass flow controller molecular weight of species i (g/kg) mass of species i (g) natural gas normal temperature and pressure (300 K and 1 bar) oxygen-enhanced combustion thermal input power of fuel (Mcal/h or kW) atmospheric pressure (bar) pressure difference between the inside of furnace and the atmosphere (mm H2O) Reynolds averaged NaviereStokes
The variation of the combustor wall divergence is known to influence the flow stream and recirculation zone in the combustor. Tu et al. [9] investigated the geometry effect of the wall divergence on the performance of a swirl stabilized burner. They reported that OEC is one of useful methods for designing a compact combustor. Previous studies in the field of OEC are summarized in Table 1 [6e10]. In the current study, the performance of an oxygen-enhanced combustor was numerically simulated and experimentally investigated to develop a pilot-scale scrap preheating furnace. An inverse diffusion type of multi-nozzle burner was used for a non-premixed oxygen-enhanced natural gas flame. The objective of the present work was to study the thermal characteristics of the pilot-scale oxygen-enhanced combustor and to optimize the effectiveness of using working fluid for scrap preheating. 2. Experimental methods A pilot-scale oxygen-enhanced combustor was developed to simulate the scrap preheating process in an EAF. Fig. 1(a) shows the geometry and dimension of a pilot-scale EAF simulator. The flow direction of the burned gas is marked by a red (in the web version) arrow along the exhaust pipe. The width (W), height (H), and depth (D) of the oxygen-enhanced furnace was 3.0, 1.0, and 1.0 m, respectively. The inside wall of the furnace was insulated with ceramic fiber board (FXL D-Block, ITM Co., Ltd., Kozaki, Chiba Prefecture, Japan) to minimize heat loss. The positive pressure difference between furnace inside and surrounding air (DP) was kept constant at about 5 ± 1.5 mm H2O to prevent air intrusion into the furnace. An oxygen-enhanced combustor was installed in the leftside wall of the furnace. The pilot-scale burner used in the current study was designed as an inverse-diffusion type with multiple nozzles. An oxygen nozzle was located in the center of the burner, while 8 fuel jet nozzles were located around the oxygen nozzle. Eight air nozzles were
RKE r SIMPLE SKE TC#n TAd Ti Tm T∞ uA uF uOx u* Vi Xi Xi,st x yþ y* z 3c
mi fG ri
realizable ke3 model transverse distance (mm) semi-implicit method for pressure linked equations standard ke3 model n-th thermocouple (ea.) adiabatic flame temperature of an oxygen-enhanced flame (K or C) temperature of species i ( C or K) volume-averaged temperature in a furnace ( C or K) surrounding temperature ( C or K) velocity in an air nozzle exit (m/s) velocity in a fuel jet nozzle exit (m/s) velocity in an oxygen nozzle exit (m/s) friction velocity volumetric flow rate of species i (lpm or Nm3/h) volumetric mole fraction of species i (%) volumetric mole fraction of species i under stoichiometric conditions (%) streamwise distance (mm) y-plus y-star (¼(mi/ri) (yþ/u*)) vertical distance (mm) combustion efficiency (%) viscosity of species i (kg/(m s)) global equivalence ratio density of species i (kg/m3)
located around the fuel nozzle to reduce the fluctuation of the oxygen-enhanced flame base. The diameters of fuel jet, oxygen, and air nozzle exits were dF ¼ 11.5 mm, dOx ¼ 18.2 mm, and dA ¼ 25.0 mm, as shown in Fig. 1(b). Gas-phase oxygen (99.0% purity; Daesung industrial gases Co., Daejeon, Korea) was supplied through a delivery pipe and an aircooled evaporator connected to a liquefied oxygen tank. The composition of natural gas (NG) was 91.6% methane (CH4), 5.8% ethane (C2H6), 1.7% propane (C3H8), 0.8% butane (C4H10), and 0.1% nitrogen (N2) (Chungnam City Gas Co., Daejeon, Korea). The composition of the surrounding air was assumed to be 21% oxygen (O2) and 79% N2 by volume. The mass flow rate of the NG, O2, and air was regulated with orifices (SOPeSOF; Samil Industry Co., Incheon, Korea), a control valve (VM-1100; Seojeon VALMAC Co., Daejeon, Korea), and mass flow controllers (5851E; Brooks Instrument Co., Hatfield, PA, USA). The mass flow controller (MFC) was calibrated using a dry gas meter (DA-16A-T; Sinagawa Co., Tokyo, Japan) before installation. The linearity of a flow rate control system was over 99.0% in 50e500 lpm in a manual. The volume-averaged temperature was measured with R-type thermocouples (TCs). These TCs were equipped on a side wall of the furnace, as shown in Fig. 1(a). The location of the TC junction (i.e. temperature measuring point) extruded 100 mm toward the inside of the furnace from the wall surface. The diameter of thermocouple junction at a thermocouple tip was about 500 mm. The radiation heat loss at the junction point, the conductive heat loss along the thermocouple wires, and the catalytic reaction on the surface of the thermocouple wires were neglected in the current study. The tolerance of TCs was estimated about 0.25% in the range of 0e1450 C in a specification sheet. A gas analyzer for nitrogen monoxide (NO) (Ultramat 6; Siemens Co., Munich, Germany), carbon monoxide (CO) (AO2020; ABB Co., Zurich, Swiss), and CO2 (Ultramat 32; Siemens Co., Munich, Germany) was used to measure the level of pollutant emissions in the
J. Oh et al. / Applied Thermal Engineering 91 (2015) 749e758
751
Table 1 Previous studies of the development of an oxygen-enhanced combustor. Authors (year) [Ref.]
Reactants (Vol. %)
Combustor
Condition
Summary
Tan et al. (2002) [8]
(1) (2) (3) (4) (1) (2) (3)
Swirl burner (W600 H8300 mm)
(1) QTh ¼ 490 kW (2) T∞ ¼ 300 K, P∞ ¼ 1 bar
Coflow burner (W1000 H300 D300 mm)
(1) QTh ¼ 25 kW (2) VNG ¼ 36 lpm (3) fG ¼ 0.952
(1) Oxygen-enhanced combustion with CO2 recirculation was good for NOx reduction and compact combustor design. (1) An increase in XO2 from 21% to 30% led to decreases of 53.5% in the heating rate and 26.1% in the fuel consumption. (1) NO and CO emissions were 5 and 25 ppm, respectively. (2) Reference wall temperature was around 870 C. (1) Combustion efficiency and heat flux increased with an increase in XO2 due to an increased CO2 and H2O mole fraction. (1) Combustion characteristics were affected by the divergence of the combustor chamber due to the variation of recirculating flow pattern.
Wu et al. (2010) [6]
NG 21, 28% O2 0, 79% N2 0, 72% CO2 NG 21e30% O2 70e79% N2
nchez et al. (2013) [10] Sa
(1) NG (97% CH4) (2) 21e35% O2 (3) 65e79% N2
Coflow burner (W600 H600 D1350 mm)
(1) QTh ¼ 20 kW (2) uA ¼ 33e59 m/s (3) T∞ ¼ 25 C, P∞ ¼ 1 bar
Bĕlohradský et al. (2014) [7]
(1) NG (2) 21e46% O2 (3) 54e79% N2
Multi-nozzle burner (dF ¼ 2.6, 3.0 mm, dOx ¼ 2.1, 2.7 mm)
(1) QTh ¼ 750 kW (2) VOx ¼ 0e120 Nm3/h
Tu et al. (2015) [9]
(1) (2) (3) (4) (3) (1) (2) (3)
Parallel jet burner (dF ¼ 50 mm, dA ¼ 124 mm)
(1) (2) (3) (4)
QTh ¼ 580 kW uF ¼ 10 m/s uA ¼ 80 m/s TA ¼ 1573 C
Multi-nozzle burner (dF ¼ 11.5 mm, dOx ¼ 18.2 mm, dA ¼ 25.0 mm)
(1) (2) (3) (4) (5)
QTh ¼ 14e35 kW uF ¼ 5.0e11.7 m/s uOx ¼ 0.0e30.0 m/s uA ¼ 0.0e20.6 m/s T∞ ¼ 300 K, P∞ ¼ 1 bar
Current study
NG (89% CH4) 19.5% O2 59.1% N2 15.0% H2O 6.4% CO2 NG (92% CH4) 21e100% O2 0e79% N2
oxygen-enhanced combustor. The location of a measuring probe was marked in Fig. 1(a). The linearity and repeatability of the flue gas measurement system was over 99.7% and 99.9% in a test report. The measured NO concentration was normalized using the emission index. The equation for the emission index of species i (EIi) is as follows [11]:
EIi ¼
mi;e ¼ mF;b
Xi XCO þ XCO2
a
MWi MWF
(1)
where mi,e is the mass of emitted species i and mF,b is the mass of fuel burned in combustion (here, mF,b ¼ mF assumes complete combustion). Xi, XCO, and XCO2 are the mole fractions of species i, CO, and CO2, respectively. a is the number of carbon in a mole of fuel (here, a ¼ 1 for CH4). MWi and MWF are the molecular weight of species i and the fuel, respectively. 3. Numerical methods The computational fluid dynamics (CFD) technique with a Fluent program was used to simulate the internal flow field and temperature distribution in an oxygen-enhanced furnace. Combustion can be defined as fast oxidation generating light and heat, including pollutant emissions if a chemical reaction can be divided into oxidation and reduction. The CFD simulation of an oxygenenhanced flame involves thermodynamics, fluid dynamics, and chemistry. The Eulerian gas-phase equations for the conservation of mass, momentum, energy, turbulence transport, and species transport were used for the CFD simulation [12]. The fluid motion of the non-compressible turbulent viscous flow was modified with a realizable ke3 (RKE) model. The RKE model is one of the Reynolds-averaged NaviereStokes (RANS) modeling techniques that simulate the process of a turbulent energy cascade from kinetic energy to thermal energy at a molecular level. This model was developed based on the standard ke3 (SKE) model, which assumes that turbulence kinetic energy (k) decays at the rate of viscous dissipation (3 ). The RKE model is known to show superior results compared to the SKE model through its use of
turbulent eddy viscosity (mt ¼ ri Cm k2/3 ). The term of Cm in the equation of mt is the variable in the RKE model while the term of Cm in the equation of mt is the constant in the SKE model. The use of Cm in the RKE model is known to show better results when simulating rotation, separation, and recirculation flow than that in the SKE model [13]. The transport equations for turbulent kinetic energy and viscous dissipation are as follows [14]:
vðr kÞ þ V$ðr u kÞ ¼ V$ vt
m vk mþ t þ Gk þ Gb r vx Prk
3 YM (2) vðr 3 Þ v þ V$ðr u 3 Þ ¼ vt vx
C2
mþ
mt Pr3 3
2
v3 þ r C1 S 3 r vx 3
pffiffiffiffiffiffiffiffiffiffi þ C13 C33 Gb k kþ n3 (3)
where k is the kinetic energy, 3 is the eddy dissipation rate, and mt is the turbulent eddy viscosity (¼ri Cm k2/3 ). Cm is the variable that is determined in a mean velocity field (¼0.09 in an inertial sub layer near a wall of S k/3 ¼ 3.3 and 0.05 in a shear layer of S k/3 ¼ 6). S is the strain rate (¼(2 Sij Sij)0.5) and Sij is the strain rate tensor. Prk and Pr3 are the turbulent Prandtl number for k and 3 (¼1.0 and 1.3). Gk and Gb are the generation of turbulence due to the mean velocity gradient and buoyancy. YM is the dilatation dissipation term ð¼ 2 r 3 Ma2t Þ where Mat is the turbulent Mach number. C1 is the variable (¼maximum of [0.43, h/(h þ 5)]) where h ¼ S k/3 . C2 and C13 are the empirical constants (¼1.92 and 1.44, respectively). C33 is the constant due to the effect of buoyancy on 3 (¼tanhjux/urj) where ux and ur are the velocity components that are parallel and normal to gravity direction. The chemical reaction of a non-premixed oxygen-enhanced flame was calculated using an eddy-dissipation concept (EDC) model with a refined two-step chemical reaction model [12]. The
752
J. Oh et al. / Applied Thermal Engineering 91 (2015) 749e758
grid were generated to consider the shear layer effect on a furnace wall. The flow velocity was set to 3 m/s in a scrap preheating section. The radial distribution of the velocity magnitude of numerical results at TC#1.5 in Fig. 1(a) was compared to that of experimental results in uF ¼ 0 m/s, uOx ¼ 0 m/a, and uA ¼ 10 m/s (i.e. VA ¼ 141.3 Nm3/h) in Fig. 3. The location of TC#1.5 was x ¼ 850 mm, y ¼ 500 mm, and z ¼ 0 mm where the origin of coordinate was the center of an oxygen nozzle exit as marked in Fig. 1(a) and (b). Mean velocity profile was measured to validate the result of numerical prediction using an anemometer (9545; TSI inc., Shoreview, MN, USA). Measured mean velocity was maximized on the centerline (i.e. x ¼ 850 mm, y ¼ 0 mm, and z ¼ 0 mm). The mean velocity profile was well matched between numerical simulation and experimental measurements as shown in Fig. 3. As boundary conditions, surrounding temperature (T∞) and pressure (P∞) were 300 K and 1 bar. The velocity in the fuel jet, oxygen, and air nozzle exits was set to uF ¼ 10 m/s, uOx ¼ 0.0 m/s, and uA ¼ 15.5 or 22.0 m/s depending on simulation conditions (i.e. Cases 1 to 2). Outlet temperature and pressure conditions were set to Tout ¼ 1000 C and DP ¼ 5 mm H2O to modify the atmosphere of hot exhaust gases and the positive pressure difference between furnace inside and surrounding air. The test conditions for the numerical simulation are listed in Table 2.
4. Results and discussion 4.1. Calculation of thermal power A pilot-scale oxygen-enhanced combustor was developed to use in the scrap preheating simulator after referring to a lab-scale burner. The characteristics and performance of a lab-scale burner have been reported in a previous study [16]. The required thermal input power (QTh) for scrap preheating was calculated as follows: T2 ¼600 Z oC
DQsteel;ideal ¼
cp;steel msteel dTsteel T1 ¼20 C
zcp;steel at 20 C msteel dTsteel Fig. 1. Geometry and dimension of (a) a pilot-scale EAF simulator and (b) oxygenenhanced combustor (unit: mm); the flow direction of burned gas in the current experiments is indicated by red arrows. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
EDC model is the extension of an eddy-dissipation model (EDM) for applying detailed chemistry based on Arrhenius-type chemical kinetics. The EDM assumes the fast chemistry and that the chemical reaction rate is governed by turbulent mixing (i.e. a large eddymixing time scale, k/3 ). Radiation heat flux was calculated with a discrete ordinate (DO) model. The DO model is suitable for various optical thicknesses of gases. The emissivity from the combustor walls and gases was set to 0.8 and 0.005, respectively. The technique of a semi-implicit method for pressure linked equations (SIMPLE) was used to calculate pressureevelocity coupling. Meanwhile, the governing transport equations were solved using second-upwind methods for spatial discretization as a numerical scheme. The analysis of grid sensitivity was performed with differentsize unstructured meshes in the rage of 2 106 to 8 106. 3,030,265 cells were selected to simulate cold and hot flow fields in an oxygen-enhanced combustor as shown in Fig. 2. In the current study, y-plus (yþ) was 200 (i.e. y* ¼ 10.9 mm) [15]. Three layers of a
(4)
¼ 0:11 100 580 ¼ 6380 kcal where DQsteel,ideal is the ideal thermal power for steel heating from T1 to T2. cp,steel is the heat capacity of steel (¼0.11 kcal/(kg K)), assumed to remain constant from 20 to 600 C). msteel is the mass of steel for heating (¼100 kg-steel). dTsteel is the volume-averaged temperature of steel (¼580 K). If considering heat loss (DQsteel,loss), and then DQsteel,real ¼ DQsteel,ideal þ DQsteel,loss z 9780 kcal where DQsteel,real is the thermal power for steel heating from T1 to T2 after considering heat loss (DQsteel,loss, assuming a heat loss of 34% during the heat transfer as usual experimental estimation in the field of a reheating furnace). When complete combustion,
DQsteel;real ¼ LHVNG Vol:NG … Vol:NG ¼
DQsteel;real 9780 z1 Nm3 ¼ 9393 LHVNG
(5)
where LHVNG is the low heating value of NG (¼9393 Kcal/Nm3). Vol.NG is the volume of NG required for scrap preheating. If the scraps of 100 kg-steel are heated for 5 min, and then the volumetric flow rate needed to heat up is 200 lpm-NG.
J. Oh et al. / Applied Thermal Engineering 91 (2015) 749e758
753
Table 2 Numerical conditions. Variable
Unit
Case 1
uF VF uOx uA XO2
m/s lpm m/s m/s %
fG
Boundary condition
2
10.0 500 0 22.0 15.5 21 0.92 1.30 TF ¼ TOx ¼ TA ¼ 300 K, Tw ¼ 1000 K, P∞ ¼ 1 bar, DP ¼ 5 mm H2O
active radical (i.e. CO þ OH / CO2 þ H). The quasi-global mechanism of CO and H2 production is as follows [18]:
Fig. 2. Grid generation of a pilot-scale EAF simulator.
Fig. 4 shows the tendency of the volumetric mole fraction of combustion products in oxygen-enhanced conditions ði:e: XO2 ¼ 40%Þ to increase the global equivalence ratio. A theoretical estimation as a non-dimensional calculation was performed in constant pressure and isothermal conditions using a Gaseq program [17]. The composition of species using the program was calculated under steady-state (i.e. equilibrium) conditions. The results are thought to be useful for estimating major or minor species in an oxygen-enhanced flame even though the reactant condition assumes as a well-stirred mixture. In Fig. 4, it can be seen that hydrogen (H2) and carbon monoxide (CO), as major species, increased with an increase in the global equivalence ratio ðfG ¼ ðXCH4 =XO2 Þ=ðXCH4 ;st =XO2 ;st ÞÞ: Here, Xi,st is the volumetric mole fraction of species i under stoichiometric conditions ði:e: XCH4 ;st =XO2 ;st ¼ 0:5Þ: The mole fraction of H2 and CO was maximized to about XH2 ¼ 28:5% and XCO ¼ 22.3% in fG ¼ 2.0. This is thought to be the result of the thermal decomposition of methane (i.e. pyrolysis). Partial oxidation in fuel-rich conditions leads to incomplete combustion because the chain reaction of CO is terminated by OH as an
Fig. 3. Verification of the mean velocity profile at TC#1.5, marked in Fig. 1(a), between numerical simulation and experimental measurement in uF ¼ 0 m/s, uOx ¼ 0 m/s, and uA ¼ 10 m/s.
CH4 / 0.5C2H4 þ H2
(R1)
C2H4 / 2CO þ 2H2
(R2)
The mole fraction of OH, atomic hydrogen (H), and nitrogen monoxide (NO), as minor species, was maximized at about XOH ¼ 4.4% in fG ¼ 0.9, XH ¼ 2.5% in fG ¼ 1.3, and XNO ¼ 1.8% in fG ¼ 0.7, respectively. OH radicals are known to be markers of the heat release rate because OH is broadly involved in the chemical reaction steps in CH4 oxidation. Oh and Noh (2015) [19] reported that the chemiluminescence intensity from OH radicals in a nonpremixed oxy-CH4 flame was much stronger than those in a CH4air flame. 4.2. Simulation of the internal flow field Figs. 5 and 6 show the distribution of velocity magnitude (a) and total temperature (b) in a pilot-scale oxygen-enhanced combustor. The geometry and dimensions are the same as those in Fig. 1. CH4 was used as a fuel instead of NG ði:e: LHVCH4 ¼ 9550 kcal=Nm3 Þ in the current simulation. The inlet velocity of air and the global equivalence ratio were varied in uA ¼ 15.5 or 22.0 m/s, and fG ¼ 0.92 or 1.30, respectively. Meanwhile, the volumetric flow rate
Fig. 4. Volumetric mole fraction of the products of OEC in relation to the global equivalence ratio in XO2 ¼ 40% (i.e. a theoretical estimation using a Gaseq program [17]).
754
J. Oh et al. / Applied Thermal Engineering 91 (2015) 749e758
Fig. 5. Internal flow field in Case 1 (i.e. QTh ¼ 286 Mcal/h, fG ¼ 0.92, and XO2 ¼ 21%Þ : the distribution of (a) velocity (utot) and (b) temperature (Ttot); the flow direction of burned gas in the current simulation is indicated by arrows.
in the fuel jet nozzle exit was fixed at VF ¼ 500 lpm (i.e. QTh ¼ 286 Mcal/h), which is the same as the initial conditions in Table 2. The flow direction of the burned gas in the simulation is indicated by the arrows. The mole fraction of oxygen in an oxidant ðXO2 Þ was defined as the percentage of the total amount of oxygen in the supplying air and pure oxygen. The equation for XO2 is expressed as follows:
XO2 ¼
VO2 ;tot total amount of oxygen ¼ total amount of oxidant VO2 þ VA
(6)
where VO2 ;tot is the total flow rate of oxygen in an oxidant. VO2 and VA are the volumetric flow rates of supplying oxygen and air, respectively. The composition of air in normal temperature and pressure (NTP) was assumed to be 21% O2 and 79% N2.
In general, near stoichiometric conditions in fG ¼ 0.92 are preferred as standard operation conditions at real work-sites because 10% of excess O2 is useful to achieve complete combustion and maximize the burned gas temperature (i.e. the furnace inside temperature). Meanwhile, fuel-rich conditions are helpful for a scale-free combustion in a reheating furnace. In general, excess O2 in a furnace leads to the oxidation of iron surface and results in the formation of metal oxide film on the iron surface compared to a stoichiometric condition at high temperatures (i.e. 1000 C). The metal oxide layer should be removed before the manufacturing process and treated as the mass loss of a basic material. In Figs. 5 and 6, it can be seen that the velocity magnitude was maximized in the centerline of a burner. The gas that was burned at a high temperature reached the other side wall due to the central jet.
J. Oh et al. / Applied Thermal Engineering 91 (2015) 749e758
755
Fig. 6. Internal flow field in Case 2 (i.e. QTh ¼ 286 Mcal/h, fG ¼ 1.30, and XO2 ¼ 21%Þ : the distribution of (a) velocity (utot) and (b) temperature (Ttot); the flow direction of burned gas in the current simulation is indicated by arrows.
The surrounding temperature in a scrap preheating section reached 1000 C in steady-state conditions when the average heat transfer coefficient of furnace wall was assumed to be 5 W/(m2 K) [20]. Hot burned gas injected from an oxygen-enhanced burner was well-stirred in the furnace and heat-up the scrap preheat zone homogeneously. The oxygen-enhanced combustion is thought to be needed for furnace and scrap heating due to the effective use of thermal energy and safety. 4.3. Performance test The performance of a pilot-scale oxygen-enhanced combustor for scrap preheating was estimated as the trend of the furnace
inside temperature with varying flow conditions. The furnace temperature (Tm) was arithmetic mean temperature which was taken from TC#1 to 6 as marked in Fig. 1(a). The experimental conditions for the performance test are listed in Table 3. Fig. 7 shows the tendencies of the temperature rising curves in constant thermal load conditions (i.e. QTh ¼ 210 Mcal/h in Cases 3, 5, 6, and 7). XO2 varied from 21% for an NGeair flame to 70% for an oxygen-enhanced flame. The characteristic time for furnace heatup until Tm ¼ 1250 C decreased as increasing XO2 : The characteristic temperature for furnace heat-up was selected as Tm ¼ 1250 C because this condition is for the thermal treatment of steel products, such as ingots, in a heating furnace in general. The
756
J. Oh et al. / Applied Thermal Engineering 91 (2015) 749e758
Table 3 Experimental conditions. Variable
QTh Vtot XO2
fG
Unit
Mcal/h Nm3/h %
Case 3
4
5
6
7
8
9
10
11
12
13
210 337 21 0.72
210 88 100 0.72
210 117 70 0.72
210 155 50 0.72
210 267 27 0.72
140 103 50 0.72
140 82 50 0.95
140 59 100 0.72
210 188 40 0.72
300 196 40 1.03
120 152 35 0.50
Fig. 7. Comparison of the temperature increase curves between CH4-air combustion and oxygen-enhanced combustion in QTh ¼ 210 Mcal/h (i.e. Cases 3, 5, 6, and 7).
characteristic time at XO2 ¼ 70% was 4.2 times shorter than that in an NGeair flame. Reducing warm-up and heating times until Tm ¼ 1250 C means an increase in productivity at a work site. This is one of the reasons for using oxy-fuel or oxygen-enhanced flames instead of an airused flame. Other benefits of OEC are increased energy savings and the reduction of CO2 emissions as well as decreased work time and fuel consumption. Fig. 8 is a comparison of the furnace heat-up patterns of Cases 6 and 7 (i.e. an experimental result), when the thermal input power varied from QTh ¼ 140 to 210 Mcal/h in XO2 ¼ 40% as shown in Table 3. The characteristic time for furnace heat-up to Tm ¼ 1250 C decreased from t ¼ 210 to 86 min as increasing QTh in a fixed amount of O2. The characteristic time in QTh ¼ 210 Mcal/h was 2.4 times shorter than that in QTh ¼ 140 Mcal/h when the thermal input power was reduced by 30%. The relationship of the furnace heat-up time with the thermal input power was not proportional even when considering different fG conditions. This non-linear trend is thought to be due to the variation of heat capacity depending on the components of exhaust gas.
Fig. 8. Comparison of the temperature increase curves between CH4-air combustion and oxygen-enhanced combustion in XO2 ¼ 40% (i.e. Cases 6 and 9).
Fig. 9 shows the relationship between the characteristic furnace heating time (t1250 C) and thermal input power (QTh). The heating time is defined as the time for furnace heat-up from Tm ¼ T∞ ¼ 25e1250 C. In Fig. 8, cases for oxygen-enhanced flames are marked as a white rectangular (i.e. ,) while cases for oxy-fuel flames are marked with a solid circle (i.e. ). As can be seen, the heating time decreased exponentially with an increase in thermal input power. The correlation between thermal input power and heating time was not linear, but rather was logarithmic. This is thought to be related to the variation of specific heat or heat capacity of burned gas because the specific heat capacity of species i (cp,i) is inversely proportional to the temperature of species i (i.e. cp,i ~ 1/Ti). Fig. 10(a) shows the pollutant emission trends from an oxygenenhanced combustor. The pollutant level of NO was normalized using the emission index of NO (EINO). The EINO increased with an increase in fG ¼ 0.72 to 0.95 while it decreased with an increase in fG ¼ 0.95 to 1.03. This trend is thought to be reasonable in general because thermal NO is dominant in high temperature conditions (over 1300 C) and burned gas temperature increases in near stoichiometric conditions (approximately fG ¼ 1.0). EINO was emitted in oxy-fuel conditions which is marked as a solid circle (i.e. ) in Fig. 9(a) even though the emission level was lower than that in oxygen-enhanced conditions. This is thought to be due to the oxidation of a small amount of N2 in NG (i.e. 0.1% N2) in pure oxygen conditions. Even a small amount of N2 addition in an oxy-fuel flame can lead to emit high level of nitrogen oxides because the adiabatic temperature of an oxy-CH4 flame is higher about 800 C than that of a CH4-air flame [21]. Fig. 10(b) shows the combustion efficiency trend in an oxygenenhanced combustor. The equation for combustion efficiency (3 c) was defined as follows [6];
3c
¼
XCO2 at 15% O2 100% XCO2 at 15% O2 þ XCO at 15% O2
(7)
where Xi at 15% O2 is the mole fraction of species i normalized at XO2 ¼ 15%: The Xi at 15% O2 was defined as follow;
Fig. 9. Performance curve of the pilot-scale oxygen-enhanced combustor when varying the furnace heating time to Ttot ¼ 1250 C in Cases 3e13.
J. Oh et al. / Applied Thermal Engineering 91 (2015) 749e758
757
5. Conclusion In the current study, the performance of a pilot-scale oxygenenhanced furnace for scrap preheating was numerically and experimentally investigated with a multi-nozzle inverse diffusion burner. NG, composed of 91.6% CH4, 5.8% C2H6, 1.7% C3H8, 0.8% C4H10, and 0.1% N2, was used as a fuel. Oxygen was diluted with air in the rage of XO2 ¼ 21%e100%: Thermal input power varied in QTh ¼ 120e300 Mcal/h. The conclusions are summarized as follows: (1) Hot burned gas injected from an oxygen-enhanced burner was well-stirred in a furnace and heat-up the scrap preheat zone homogeneously. The volume-averaged temperature in the furnace was predicted at about 1250 C in steady state conditions. (2) The tendency of temperature rising in the furnace increased steeply with an increase in the oxygen mole fraction in an oxidant. The characteristic time for furnace heat-up increased exponentially as the thermal input power increased. The correlation between thermal input power and heating time was logarithmic, not linear. (3) The EINO increased with an increase in fG ¼ 0.72 to 0.95 while EINO decreased with an increase in fG ¼ 0.95 to 1.03. Even when a small amount of N2 was added to an oxy-fuel flame, this led to a high level of nitrogen oxides in comparison with an oxygen-enhanced flame. Acknowledgements This work was supported by an Energy Efficiency & Resources Core Technology Program of the Korea Institute of Energy Technology Evaluation and Planning (KETEP) grant funded by the Korean Government Ministry of Knowledge Economy (NP20130026-3). References
Fig. 10. Performance of a pilot-scale oxygen-enhanced combustor; (a) pollutant emission level and (b) combustion efficiency under the varying operating conditions in Table 3.
Xi at 15% O2 ¼ Xi
XO2 ;in 15% 100% XO2 ;in XO2 ;out
(8)
where Xi is the mole fraction of species i, XO2 ;in and XO2 ;out are the mole fraction of O2 in an oxidant and flue gas, respectively. The combustion efficiency of non-premixed oxygen-enhanced flames was maximized as 99.4% in fG ¼ 0.72 for both cases of oxyNG combustion and OEC. Meanwhile, the combustion efficiency in other cases was 62.7e81.5 in fG ¼ 0.50, 0.95, and 1.03. An oxygenenhanced flame in fG ¼ 0.95 and 1.03 is considered to be locally incomplete combustion even though the global equivalence is near stoichiometric conditions. In the case of fG ¼ 0.50, the oxygenenhanced flame is thought to be near blow-out region because EINO was maximized even in fuel-lean conditions. This means that a locally fuel-rich flame base was stabilized near the nozzle exit while local flame extinction occurred around the flame tip region due to the highly stretched flow.
[1] L. Zheng, Oxy-fuel Combustion for Power Generation and Carbon Dioxide (CO2) Capture, Woodhead Publishing Limited, Cambridge, UK, 2011. [2] Y.N. Toulouevski, I.Y. Zinurov, Innovation in Electric Arc Furnaces, SpringerVerlag, Berlin, Germany, 2010. [3] C.E. Baukal, Oxygen-enhanced Combustion, CRC Press, Boca Raton, USA, 2000. [4] Website address: https://steelplantech.com/product/ecoarc/. [5] Website address: http://www.corefurnace.com/meltshop_01.html. [6] K.K. Wu, Y.C. Chang, C.H. Chen, Y.D. Chen, High-efficiency combustion of natural gas with 21e30% oxygen-enriched air, Fuel 89 (2010) 2455e2462. k, Experimental study on the influence of [7] P. Bĕlohradský, P. Skryja, I. Huda oxygen content in the combustion air on the combustion characteristics, Energy 75 (2014) 116e126. [8] Y. Tan, M.A. Douglas, K.V. Thambimuthu, CO2 capturing using oxygen enhanced combustion strategies for natural gas power plants, Fuel 81 (2002) 1007e1016. [9] Y. Tu, H. Liu, S. Chen, Z. Liu, H. Zhao, C. Zheng, Effects of furnace chamber shape on the MILD combustion of natural gas, Appl. Therm. Eng. 76 (2015) 64e75. nchez, F. Cadavid, A. Amell, Experimental evaluation of a 20 kW oxygen [10] M. Sa enhanced self-regenerative burner operated in flameless combustion mode, Appl. Energy 111 (2013) 240e246. [11] S. Turns, An Introduction to Combustion: Concepts and Applications, third ed., McGraw-Hill, New York, USA, 2012, pp. 559e560. [12] J. Oh, S. Dong, J. Yang, The effect of fuel composition on the characteristics of a non-premixed synthetic natural gas-air flame, Fuel Process. Technol. 126 (2014) 66e75. [13] ANSYS FLUENT 12.0 Theory Guide, ANSYS Inc, Canonsburg, USA, 2009. [14] T.H. Shih, W.W. Liou, A. Shabbir, Z. Yang, J. Zhu, A new ke3 eddy-viscosity model for high Reynolds number turbulent flows - model development and validation, Comput. Fluids 24 (3) (1995) 227e238. [15] F.M. White, Fluid Mechanics, seventh ed., vol. 362e364, McGraw Hill, New York, NY, USA, 2011, pp. 471e473. [16] J.I. Seo, Y.T. Guahk, S.H. Bae, J.G. Hong, U.D. Lee, H.D. Shin, Experimental study on the oxygen combustion characteristics with CO2 feeding, J. Korean Soc. Combust. 20 (2) (2005) 26e34. [17] Website address: http://www.gaseq.co.uk/.
758
J. Oh et al. / Applied Thermal Engineering 91 (2015) 749e758
[18] D.J. Hautman, F.L. Dryer, K.P. Schug, I. Glassman, A multiple-step over all kinetic mechanism for the oxidation of hydrocarbons, Combust. Sci. Technol. 25 (1981) 219e235. [19] J. Oh, D. Noh, Flame characteristics of a non-premixed oxy-fuel jet in a labscale furnace, Energy 81 (2015) 328e343.
[20] A. Abbassi, Kh. Khoshmanesh, Numerical simulation and experimental analysis of an industrial glass melting furnace, Appl. Therm. Eng. 28 (2008) 450e459. [21] J. Oh, D. Noh, Laminar burning velocity of oxy-methane flames in atmospheric condition, Energy 45 (2012) 669e675.
Contents lists available at ScienceDirect
Applied Thermal Engineering journal homepage: www.elsevier.com/locate/apthermeng
Research paper
Development of an oxygen-enhanced combustor for scrap preheating in an electric arc furnace Jeongseog Oh*, Eungyeong Lee, Dongsoon Noh Advanced Combustion Laboratory, Korea Institute of Energy Research, Daejeon, Republic of Korea
a r t i c l e i n f o
a b s t r a c t
Article history: Received 25 June 2015 Accepted 30 August 2015 Available online 5 September 2015
The performance of a pilot-scale oxygen-enhanced furnace for scrap preheating with a multi-nozzle inverse diffusion combustor was studied. Natural gas and oxygen diluted in air were supplied to the oxygen-enhanced combustor with a range of 120e300 Mcal/h. The internal flow field of the furnace was simulated with a computational fluid dynamics program before any experiments were carried out. Meanwhile, the characteristic time for furnace heat-up and the level of pollutant emission were experimentally measured. Results showed that the temperature rise in the furnace increased steeply with an increase in the oxygen mole fraction in an oxidant. In addition, the characteristic time for furnace heat-up decreased exponentially as the thermal input power increased. © 2015 Elsevier Ltd. All rights reserved.
Keywords: Oxy-fuel combustion Oxygen-enhancement CH4-air flame Scrap preheating Electric arc furnace
1. Introduction Air-used combustion is a method that is commonly used in the steel melting or reheating processes. In recent, oxygen-enhanced or oxygen-enriched combustion (OEC) is introduced to increase thermal efficiency [1]. Steel and iron have served as basic materials for industry. The steel making process is classified as either blast furnace or electric arc furnace (EAF) method. Compared to the blast furnace, on advantage of the EAF method is increased energy savings due to the simple process and the low emission of carbon dioxide (CO2) [2]. In particular, this is relevant today because of the world wide attempts to reduce CO2 emissions due to global warming and climate change [3]. A new method for an effective thermochemical steel-making using in EAF process is needed to reduce energy consumption. Several alternatives that have been suggested use heat recovery from burned gas (i.e. flue gas), such as the Eco-Arc and Consteel processes [4,5]. The core concept of both methods is scrap preheating up to 600 C with high temperature flue gas (about 1200 C). As a secondary thermal energy source into relatively cold zone, OEC and carbon lancing play an important role in increasing electric energy savings. This thermochemical method is one of
* Corresponding author. Advanced Combustion Laboratory, Korea Institute of Energy Research, Daejeon 305-343, Republic of Korea. Tel.: þ82 42 860 3479; fax: þ82 42 860 3133. E-mail address: [email protected] (J. Oh). http://dx.doi.org/10.1016/j.applthermaleng.2015.08.088 1359-4311/© 2015 Elsevier Ltd. All rights reserved.
ways to reduce electricity usage by removing the energy conversion process of fossil fuel energy to electric energy. OEC is characterized by a higher flame temperature than that of air-used combustion because a higher oxygen concentration in an oxidant leads to a lower absorption energy by nitrogen (N2). This characteristic of the high temperature has been used in heat treatment furnaces, waste incinerators, propulsion engines, etc. Compared to air-used combustion, OEC is known to increase reaction rate due to the higher flame temperature and oxygen mole fraction in the same flow condition. Wu et al. [6] reported that a small amount of oxygen added to an oxidant (e.g. 21%e30%) resulted in a fuel consumption decrease of 26.1%. Increased OEC is good for not only energy savings, but also for productivity improvement. Bĕlohradský et al. [7] reported that combustion efficiency is influenced by the mole fraction of O2 in an oxidant ðXO2 Þ due to an increased CO2 and water vapor (H2O) mole fraction in flue gas. A flue gas temperature is in inverse proportion to the time needed for the heat transfer to the object being heated, such as iron nuggets and ingots. However, a weakness of OEC is the additional costs related to the increased oxygen supply and burner retrofitting because the increased combustion intensity leads to a high temperature environment in the furnace and the exposure of the burner surface to this heat level results in a need for more frequent maintenance. A method using CO2 recirculation has been suggested to control reaction rate and flame temperature [8]. CO2 dilution to an oxidant is effective in reducing the production of thermal nitrogen oxides (NOx).
750
J. Oh et al. / Applied Thermal Engineering 91 (2015) 749e758
Nomenclature a.u. CCS cpi DO dA dF dOx EAF EDM EDC EIi LHVi lpm MFC MWi mi NG NTP OEC QTh P∞ DP RANS
arbitrary unit carbon capture and storage specific heat of species i (cal/(g K)) discrete ordinate model outlet diameter in an air nozzle exit (mm) outlet diameter in a fuel jet nozzle exit (mm) outlet diameter in an oxygen nozzle exit (mm) electric arc furnace eddy-dissipation model eddy-dissipation concept model emission index of species i (g/kg) low heating value of species i (kcal/Nm3) liter per minute mass flow controller molecular weight of species i (g/kg) mass of species i (g) natural gas normal temperature and pressure (300 K and 1 bar) oxygen-enhanced combustion thermal input power of fuel (Mcal/h or kW) atmospheric pressure (bar) pressure difference between the inside of furnace and the atmosphere (mm H2O) Reynolds averaged NaviereStokes
The variation of the combustor wall divergence is known to influence the flow stream and recirculation zone in the combustor. Tu et al. [9] investigated the geometry effect of the wall divergence on the performance of a swirl stabilized burner. They reported that OEC is one of useful methods for designing a compact combustor. Previous studies in the field of OEC are summarized in Table 1 [6e10]. In the current study, the performance of an oxygen-enhanced combustor was numerically simulated and experimentally investigated to develop a pilot-scale scrap preheating furnace. An inverse diffusion type of multi-nozzle burner was used for a non-premixed oxygen-enhanced natural gas flame. The objective of the present work was to study the thermal characteristics of the pilot-scale oxygen-enhanced combustor and to optimize the effectiveness of using working fluid for scrap preheating. 2. Experimental methods A pilot-scale oxygen-enhanced combustor was developed to simulate the scrap preheating process in an EAF. Fig. 1(a) shows the geometry and dimension of a pilot-scale EAF simulator. The flow direction of the burned gas is marked by a red (in the web version) arrow along the exhaust pipe. The width (W), height (H), and depth (D) of the oxygen-enhanced furnace was 3.0, 1.0, and 1.0 m, respectively. The inside wall of the furnace was insulated with ceramic fiber board (FXL D-Block, ITM Co., Ltd., Kozaki, Chiba Prefecture, Japan) to minimize heat loss. The positive pressure difference between furnace inside and surrounding air (DP) was kept constant at about 5 ± 1.5 mm H2O to prevent air intrusion into the furnace. An oxygen-enhanced combustor was installed in the leftside wall of the furnace. The pilot-scale burner used in the current study was designed as an inverse-diffusion type with multiple nozzles. An oxygen nozzle was located in the center of the burner, while 8 fuel jet nozzles were located around the oxygen nozzle. Eight air nozzles were
RKE r SIMPLE SKE TC#n TAd Ti Tm T∞ uA uF uOx u* Vi Xi Xi,st x yþ y* z 3c
mi fG ri
realizable ke3 model transverse distance (mm) semi-implicit method for pressure linked equations standard ke3 model n-th thermocouple (ea.) adiabatic flame temperature of an oxygen-enhanced flame (K or C) temperature of species i ( C or K) volume-averaged temperature in a furnace ( C or K) surrounding temperature ( C or K) velocity in an air nozzle exit (m/s) velocity in a fuel jet nozzle exit (m/s) velocity in an oxygen nozzle exit (m/s) friction velocity volumetric flow rate of species i (lpm or Nm3/h) volumetric mole fraction of species i (%) volumetric mole fraction of species i under stoichiometric conditions (%) streamwise distance (mm) y-plus y-star (¼(mi/ri) (yþ/u*)) vertical distance (mm) combustion efficiency (%) viscosity of species i (kg/(m s)) global equivalence ratio density of species i (kg/m3)
located around the fuel nozzle to reduce the fluctuation of the oxygen-enhanced flame base. The diameters of fuel jet, oxygen, and air nozzle exits were dF ¼ 11.5 mm, dOx ¼ 18.2 mm, and dA ¼ 25.0 mm, as shown in Fig. 1(b). Gas-phase oxygen (99.0% purity; Daesung industrial gases Co., Daejeon, Korea) was supplied through a delivery pipe and an aircooled evaporator connected to a liquefied oxygen tank. The composition of natural gas (NG) was 91.6% methane (CH4), 5.8% ethane (C2H6), 1.7% propane (C3H8), 0.8% butane (C4H10), and 0.1% nitrogen (N2) (Chungnam City Gas Co., Daejeon, Korea). The composition of the surrounding air was assumed to be 21% oxygen (O2) and 79% N2 by volume. The mass flow rate of the NG, O2, and air was regulated with orifices (SOPeSOF; Samil Industry Co., Incheon, Korea), a control valve (VM-1100; Seojeon VALMAC Co., Daejeon, Korea), and mass flow controllers (5851E; Brooks Instrument Co., Hatfield, PA, USA). The mass flow controller (MFC) was calibrated using a dry gas meter (DA-16A-T; Sinagawa Co., Tokyo, Japan) before installation. The linearity of a flow rate control system was over 99.0% in 50e500 lpm in a manual. The volume-averaged temperature was measured with R-type thermocouples (TCs). These TCs were equipped on a side wall of the furnace, as shown in Fig. 1(a). The location of the TC junction (i.e. temperature measuring point) extruded 100 mm toward the inside of the furnace from the wall surface. The diameter of thermocouple junction at a thermocouple tip was about 500 mm. The radiation heat loss at the junction point, the conductive heat loss along the thermocouple wires, and the catalytic reaction on the surface of the thermocouple wires were neglected in the current study. The tolerance of TCs was estimated about 0.25% in the range of 0e1450 C in a specification sheet. A gas analyzer for nitrogen monoxide (NO) (Ultramat 6; Siemens Co., Munich, Germany), carbon monoxide (CO) (AO2020; ABB Co., Zurich, Swiss), and CO2 (Ultramat 32; Siemens Co., Munich, Germany) was used to measure the level of pollutant emissions in the
J. Oh et al. / Applied Thermal Engineering 91 (2015) 749e758
751
Table 1 Previous studies of the development of an oxygen-enhanced combustor. Authors (year) [Ref.]
Reactants (Vol. %)
Combustor
Condition
Summary
Tan et al. (2002) [8]
(1) (2) (3) (4) (1) (2) (3)
Swirl burner (W600 H8300 mm)
(1) QTh ¼ 490 kW (2) T∞ ¼ 300 K, P∞ ¼ 1 bar
Coflow burner (W1000 H300 D300 mm)
(1) QTh ¼ 25 kW (2) VNG ¼ 36 lpm (3) fG ¼ 0.952
(1) Oxygen-enhanced combustion with CO2 recirculation was good for NOx reduction and compact combustor design. (1) An increase in XO2 from 21% to 30% led to decreases of 53.5% in the heating rate and 26.1% in the fuel consumption. (1) NO and CO emissions were 5 and 25 ppm, respectively. (2) Reference wall temperature was around 870 C. (1) Combustion efficiency and heat flux increased with an increase in XO2 due to an increased CO2 and H2O mole fraction. (1) Combustion characteristics were affected by the divergence of the combustor chamber due to the variation of recirculating flow pattern.
Wu et al. (2010) [6]
NG 21, 28% O2 0, 79% N2 0, 72% CO2 NG 21e30% O2 70e79% N2
nchez et al. (2013) [10] Sa
(1) NG (97% CH4) (2) 21e35% O2 (3) 65e79% N2
Coflow burner (W600 H600 D1350 mm)
(1) QTh ¼ 20 kW (2) uA ¼ 33e59 m/s (3) T∞ ¼ 25 C, P∞ ¼ 1 bar
Bĕlohradský et al. (2014) [7]
(1) NG (2) 21e46% O2 (3) 54e79% N2
Multi-nozzle burner (dF ¼ 2.6, 3.0 mm, dOx ¼ 2.1, 2.7 mm)
(1) QTh ¼ 750 kW (2) VOx ¼ 0e120 Nm3/h
Tu et al. (2015) [9]
(1) (2) (3) (4) (3) (1) (2) (3)
Parallel jet burner (dF ¼ 50 mm, dA ¼ 124 mm)
(1) (2) (3) (4)
QTh ¼ 580 kW uF ¼ 10 m/s uA ¼ 80 m/s TA ¼ 1573 C
Multi-nozzle burner (dF ¼ 11.5 mm, dOx ¼ 18.2 mm, dA ¼ 25.0 mm)
(1) (2) (3) (4) (5)
QTh ¼ 14e35 kW uF ¼ 5.0e11.7 m/s uOx ¼ 0.0e30.0 m/s uA ¼ 0.0e20.6 m/s T∞ ¼ 300 K, P∞ ¼ 1 bar
Current study
NG (89% CH4) 19.5% O2 59.1% N2 15.0% H2O 6.4% CO2 NG (92% CH4) 21e100% O2 0e79% N2
oxygen-enhanced combustor. The location of a measuring probe was marked in Fig. 1(a). The linearity and repeatability of the flue gas measurement system was over 99.7% and 99.9% in a test report. The measured NO concentration was normalized using the emission index. The equation for the emission index of species i (EIi) is as follows [11]:
EIi ¼
mi;e ¼ mF;b
Xi XCO þ XCO2
a
MWi MWF
(1)
where mi,e is the mass of emitted species i and mF,b is the mass of fuel burned in combustion (here, mF,b ¼ mF assumes complete combustion). Xi, XCO, and XCO2 are the mole fractions of species i, CO, and CO2, respectively. a is the number of carbon in a mole of fuel (here, a ¼ 1 for CH4). MWi and MWF are the molecular weight of species i and the fuel, respectively. 3. Numerical methods The computational fluid dynamics (CFD) technique with a Fluent program was used to simulate the internal flow field and temperature distribution in an oxygen-enhanced furnace. Combustion can be defined as fast oxidation generating light and heat, including pollutant emissions if a chemical reaction can be divided into oxidation and reduction. The CFD simulation of an oxygenenhanced flame involves thermodynamics, fluid dynamics, and chemistry. The Eulerian gas-phase equations for the conservation of mass, momentum, energy, turbulence transport, and species transport were used for the CFD simulation [12]. The fluid motion of the non-compressible turbulent viscous flow was modified with a realizable ke3 (RKE) model. The RKE model is one of the Reynolds-averaged NaviereStokes (RANS) modeling techniques that simulate the process of a turbulent energy cascade from kinetic energy to thermal energy at a molecular level. This model was developed based on the standard ke3 (SKE) model, which assumes that turbulence kinetic energy (k) decays at the rate of viscous dissipation (3 ). The RKE model is known to show superior results compared to the SKE model through its use of
turbulent eddy viscosity (mt ¼ ri Cm k2/3 ). The term of Cm in the equation of mt is the variable in the RKE model while the term of Cm in the equation of mt is the constant in the SKE model. The use of Cm in the RKE model is known to show better results when simulating rotation, separation, and recirculation flow than that in the SKE model [13]. The transport equations for turbulent kinetic energy and viscous dissipation are as follows [14]:
vðr kÞ þ V$ðr u kÞ ¼ V$ vt
m vk mþ t þ Gk þ Gb r vx Prk
3 YM (2) vðr 3 Þ v þ V$ðr u 3 Þ ¼ vt vx
C2
mþ
mt Pr3 3
2
v3 þ r C1 S 3 r vx 3
pffiffiffiffiffiffiffiffiffiffi þ C13 C33 Gb k kþ n3 (3)
where k is the kinetic energy, 3 is the eddy dissipation rate, and mt is the turbulent eddy viscosity (¼ri Cm k2/3 ). Cm is the variable that is determined in a mean velocity field (¼0.09 in an inertial sub layer near a wall of S k/3 ¼ 3.3 and 0.05 in a shear layer of S k/3 ¼ 6). S is the strain rate (¼(2 Sij Sij)0.5) and Sij is the strain rate tensor. Prk and Pr3 are the turbulent Prandtl number for k and 3 (¼1.0 and 1.3). Gk and Gb are the generation of turbulence due to the mean velocity gradient and buoyancy. YM is the dilatation dissipation term ð¼ 2 r 3 Ma2t Þ where Mat is the turbulent Mach number. C1 is the variable (¼maximum of [0.43, h/(h þ 5)]) where h ¼ S k/3 . C2 and C13 are the empirical constants (¼1.92 and 1.44, respectively). C33 is the constant due to the effect of buoyancy on 3 (¼tanhjux/urj) where ux and ur are the velocity components that are parallel and normal to gravity direction. The chemical reaction of a non-premixed oxygen-enhanced flame was calculated using an eddy-dissipation concept (EDC) model with a refined two-step chemical reaction model [12]. The
752
J. Oh et al. / Applied Thermal Engineering 91 (2015) 749e758
grid were generated to consider the shear layer effect on a furnace wall. The flow velocity was set to 3 m/s in a scrap preheating section. The radial distribution of the velocity magnitude of numerical results at TC#1.5 in Fig. 1(a) was compared to that of experimental results in uF ¼ 0 m/s, uOx ¼ 0 m/a, and uA ¼ 10 m/s (i.e. VA ¼ 141.3 Nm3/h) in Fig. 3. The location of TC#1.5 was x ¼ 850 mm, y ¼ 500 mm, and z ¼ 0 mm where the origin of coordinate was the center of an oxygen nozzle exit as marked in Fig. 1(a) and (b). Mean velocity profile was measured to validate the result of numerical prediction using an anemometer (9545; TSI inc., Shoreview, MN, USA). Measured mean velocity was maximized on the centerline (i.e. x ¼ 850 mm, y ¼ 0 mm, and z ¼ 0 mm). The mean velocity profile was well matched between numerical simulation and experimental measurements as shown in Fig. 3. As boundary conditions, surrounding temperature (T∞) and pressure (P∞) were 300 K and 1 bar. The velocity in the fuel jet, oxygen, and air nozzle exits was set to uF ¼ 10 m/s, uOx ¼ 0.0 m/s, and uA ¼ 15.5 or 22.0 m/s depending on simulation conditions (i.e. Cases 1 to 2). Outlet temperature and pressure conditions were set to Tout ¼ 1000 C and DP ¼ 5 mm H2O to modify the atmosphere of hot exhaust gases and the positive pressure difference between furnace inside and surrounding air. The test conditions for the numerical simulation are listed in Table 2.
4. Results and discussion 4.1. Calculation of thermal power A pilot-scale oxygen-enhanced combustor was developed to use in the scrap preheating simulator after referring to a lab-scale burner. The characteristics and performance of a lab-scale burner have been reported in a previous study [16]. The required thermal input power (QTh) for scrap preheating was calculated as follows: T2 ¼600 Z oC
DQsteel;ideal ¼
cp;steel msteel dTsteel T1 ¼20 C
zcp;steel at 20 C msteel dTsteel Fig. 1. Geometry and dimension of (a) a pilot-scale EAF simulator and (b) oxygenenhanced combustor (unit: mm); the flow direction of burned gas in the current experiments is indicated by red arrows. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
EDC model is the extension of an eddy-dissipation model (EDM) for applying detailed chemistry based on Arrhenius-type chemical kinetics. The EDM assumes the fast chemistry and that the chemical reaction rate is governed by turbulent mixing (i.e. a large eddymixing time scale, k/3 ). Radiation heat flux was calculated with a discrete ordinate (DO) model. The DO model is suitable for various optical thicknesses of gases. The emissivity from the combustor walls and gases was set to 0.8 and 0.005, respectively. The technique of a semi-implicit method for pressure linked equations (SIMPLE) was used to calculate pressureevelocity coupling. Meanwhile, the governing transport equations were solved using second-upwind methods for spatial discretization as a numerical scheme. The analysis of grid sensitivity was performed with differentsize unstructured meshes in the rage of 2 106 to 8 106. 3,030,265 cells were selected to simulate cold and hot flow fields in an oxygen-enhanced combustor as shown in Fig. 2. In the current study, y-plus (yþ) was 200 (i.e. y* ¼ 10.9 mm) [15]. Three layers of a
(4)
¼ 0:11 100 580 ¼ 6380 kcal where DQsteel,ideal is the ideal thermal power for steel heating from T1 to T2. cp,steel is the heat capacity of steel (¼0.11 kcal/(kg K)), assumed to remain constant from 20 to 600 C). msteel is the mass of steel for heating (¼100 kg-steel). dTsteel is the volume-averaged temperature of steel (¼580 K). If considering heat loss (DQsteel,loss), and then DQsteel,real ¼ DQsteel,ideal þ DQsteel,loss z 9780 kcal where DQsteel,real is the thermal power for steel heating from T1 to T2 after considering heat loss (DQsteel,loss, assuming a heat loss of 34% during the heat transfer as usual experimental estimation in the field of a reheating furnace). When complete combustion,
DQsteel;real ¼ LHVNG Vol:NG … Vol:NG ¼
DQsteel;real 9780 z1 Nm3 ¼ 9393 LHVNG
(5)
where LHVNG is the low heating value of NG (¼9393 Kcal/Nm3). Vol.NG is the volume of NG required for scrap preheating. If the scraps of 100 kg-steel are heated for 5 min, and then the volumetric flow rate needed to heat up is 200 lpm-NG.
J. Oh et al. / Applied Thermal Engineering 91 (2015) 749e758
753
Table 2 Numerical conditions. Variable
Unit
Case 1
uF VF uOx uA XO2
m/s lpm m/s m/s %
fG
Boundary condition
2
10.0 500 0 22.0 15.5 21 0.92 1.30 TF ¼ TOx ¼ TA ¼ 300 K, Tw ¼ 1000 K, P∞ ¼ 1 bar, DP ¼ 5 mm H2O
active radical (i.e. CO þ OH / CO2 þ H). The quasi-global mechanism of CO and H2 production is as follows [18]:
Fig. 2. Grid generation of a pilot-scale EAF simulator.
Fig. 4 shows the tendency of the volumetric mole fraction of combustion products in oxygen-enhanced conditions ði:e: XO2 ¼ 40%Þ to increase the global equivalence ratio. A theoretical estimation as a non-dimensional calculation was performed in constant pressure and isothermal conditions using a Gaseq program [17]. The composition of species using the program was calculated under steady-state (i.e. equilibrium) conditions. The results are thought to be useful for estimating major or minor species in an oxygen-enhanced flame even though the reactant condition assumes as a well-stirred mixture. In Fig. 4, it can be seen that hydrogen (H2) and carbon monoxide (CO), as major species, increased with an increase in the global equivalence ratio ðfG ¼ ðXCH4 =XO2 Þ=ðXCH4 ;st =XO2 ;st ÞÞ: Here, Xi,st is the volumetric mole fraction of species i under stoichiometric conditions ði:e: XCH4 ;st =XO2 ;st ¼ 0:5Þ: The mole fraction of H2 and CO was maximized to about XH2 ¼ 28:5% and XCO ¼ 22.3% in fG ¼ 2.0. This is thought to be the result of the thermal decomposition of methane (i.e. pyrolysis). Partial oxidation in fuel-rich conditions leads to incomplete combustion because the chain reaction of CO is terminated by OH as an
Fig. 3. Verification of the mean velocity profile at TC#1.5, marked in Fig. 1(a), between numerical simulation and experimental measurement in uF ¼ 0 m/s, uOx ¼ 0 m/s, and uA ¼ 10 m/s.
CH4 / 0.5C2H4 þ H2
(R1)
C2H4 / 2CO þ 2H2
(R2)
The mole fraction of OH, atomic hydrogen (H), and nitrogen monoxide (NO), as minor species, was maximized at about XOH ¼ 4.4% in fG ¼ 0.9, XH ¼ 2.5% in fG ¼ 1.3, and XNO ¼ 1.8% in fG ¼ 0.7, respectively. OH radicals are known to be markers of the heat release rate because OH is broadly involved in the chemical reaction steps in CH4 oxidation. Oh and Noh (2015) [19] reported that the chemiluminescence intensity from OH radicals in a nonpremixed oxy-CH4 flame was much stronger than those in a CH4air flame. 4.2. Simulation of the internal flow field Figs. 5 and 6 show the distribution of velocity magnitude (a) and total temperature (b) in a pilot-scale oxygen-enhanced combustor. The geometry and dimensions are the same as those in Fig. 1. CH4 was used as a fuel instead of NG ði:e: LHVCH4 ¼ 9550 kcal=Nm3 Þ in the current simulation. The inlet velocity of air and the global equivalence ratio were varied in uA ¼ 15.5 or 22.0 m/s, and fG ¼ 0.92 or 1.30, respectively. Meanwhile, the volumetric flow rate
Fig. 4. Volumetric mole fraction of the products of OEC in relation to the global equivalence ratio in XO2 ¼ 40% (i.e. a theoretical estimation using a Gaseq program [17]).
754
J. Oh et al. / Applied Thermal Engineering 91 (2015) 749e758
Fig. 5. Internal flow field in Case 1 (i.e. QTh ¼ 286 Mcal/h, fG ¼ 0.92, and XO2 ¼ 21%Þ : the distribution of (a) velocity (utot) and (b) temperature (Ttot); the flow direction of burned gas in the current simulation is indicated by arrows.
in the fuel jet nozzle exit was fixed at VF ¼ 500 lpm (i.e. QTh ¼ 286 Mcal/h), which is the same as the initial conditions in Table 2. The flow direction of the burned gas in the simulation is indicated by the arrows. The mole fraction of oxygen in an oxidant ðXO2 Þ was defined as the percentage of the total amount of oxygen in the supplying air and pure oxygen. The equation for XO2 is expressed as follows:
XO2 ¼
VO2 ;tot total amount of oxygen ¼ total amount of oxidant VO2 þ VA
(6)
where VO2 ;tot is the total flow rate of oxygen in an oxidant. VO2 and VA are the volumetric flow rates of supplying oxygen and air, respectively. The composition of air in normal temperature and pressure (NTP) was assumed to be 21% O2 and 79% N2.
In general, near stoichiometric conditions in fG ¼ 0.92 are preferred as standard operation conditions at real work-sites because 10% of excess O2 is useful to achieve complete combustion and maximize the burned gas temperature (i.e. the furnace inside temperature). Meanwhile, fuel-rich conditions are helpful for a scale-free combustion in a reheating furnace. In general, excess O2 in a furnace leads to the oxidation of iron surface and results in the formation of metal oxide film on the iron surface compared to a stoichiometric condition at high temperatures (i.e. 1000 C). The metal oxide layer should be removed before the manufacturing process and treated as the mass loss of a basic material. In Figs. 5 and 6, it can be seen that the velocity magnitude was maximized in the centerline of a burner. The gas that was burned at a high temperature reached the other side wall due to the central jet.
J. Oh et al. / Applied Thermal Engineering 91 (2015) 749e758
755
Fig. 6. Internal flow field in Case 2 (i.e. QTh ¼ 286 Mcal/h, fG ¼ 1.30, and XO2 ¼ 21%Þ : the distribution of (a) velocity (utot) and (b) temperature (Ttot); the flow direction of burned gas in the current simulation is indicated by arrows.
The surrounding temperature in a scrap preheating section reached 1000 C in steady-state conditions when the average heat transfer coefficient of furnace wall was assumed to be 5 W/(m2 K) [20]. Hot burned gas injected from an oxygen-enhanced burner was well-stirred in the furnace and heat-up the scrap preheat zone homogeneously. The oxygen-enhanced combustion is thought to be needed for furnace and scrap heating due to the effective use of thermal energy and safety. 4.3. Performance test The performance of a pilot-scale oxygen-enhanced combustor for scrap preheating was estimated as the trend of the furnace
inside temperature with varying flow conditions. The furnace temperature (Tm) was arithmetic mean temperature which was taken from TC#1 to 6 as marked in Fig. 1(a). The experimental conditions for the performance test are listed in Table 3. Fig. 7 shows the tendencies of the temperature rising curves in constant thermal load conditions (i.e. QTh ¼ 210 Mcal/h in Cases 3, 5, 6, and 7). XO2 varied from 21% for an NGeair flame to 70% for an oxygen-enhanced flame. The characteristic time for furnace heatup until Tm ¼ 1250 C decreased as increasing XO2 : The characteristic temperature for furnace heat-up was selected as Tm ¼ 1250 C because this condition is for the thermal treatment of steel products, such as ingots, in a heating furnace in general. The
756
J. Oh et al. / Applied Thermal Engineering 91 (2015) 749e758
Table 3 Experimental conditions. Variable
QTh Vtot XO2
fG
Unit
Mcal/h Nm3/h %
Case 3
4
5
6
7
8
9
10
11
12
13
210 337 21 0.72
210 88 100 0.72
210 117 70 0.72
210 155 50 0.72
210 267 27 0.72
140 103 50 0.72
140 82 50 0.95
140 59 100 0.72
210 188 40 0.72
300 196 40 1.03
120 152 35 0.50
Fig. 7. Comparison of the temperature increase curves between CH4-air combustion and oxygen-enhanced combustion in QTh ¼ 210 Mcal/h (i.e. Cases 3, 5, 6, and 7).
characteristic time at XO2 ¼ 70% was 4.2 times shorter than that in an NGeair flame. Reducing warm-up and heating times until Tm ¼ 1250 C means an increase in productivity at a work site. This is one of the reasons for using oxy-fuel or oxygen-enhanced flames instead of an airused flame. Other benefits of OEC are increased energy savings and the reduction of CO2 emissions as well as decreased work time and fuel consumption. Fig. 8 is a comparison of the furnace heat-up patterns of Cases 6 and 7 (i.e. an experimental result), when the thermal input power varied from QTh ¼ 140 to 210 Mcal/h in XO2 ¼ 40% as shown in Table 3. The characteristic time for furnace heat-up to Tm ¼ 1250 C decreased from t ¼ 210 to 86 min as increasing QTh in a fixed amount of O2. The characteristic time in QTh ¼ 210 Mcal/h was 2.4 times shorter than that in QTh ¼ 140 Mcal/h when the thermal input power was reduced by 30%. The relationship of the furnace heat-up time with the thermal input power was not proportional even when considering different fG conditions. This non-linear trend is thought to be due to the variation of heat capacity depending on the components of exhaust gas.
Fig. 8. Comparison of the temperature increase curves between CH4-air combustion and oxygen-enhanced combustion in XO2 ¼ 40% (i.e. Cases 6 and 9).
Fig. 9 shows the relationship between the characteristic furnace heating time (t1250 C) and thermal input power (QTh). The heating time is defined as the time for furnace heat-up from Tm ¼ T∞ ¼ 25e1250 C. In Fig. 8, cases for oxygen-enhanced flames are marked as a white rectangular (i.e. ,) while cases for oxy-fuel flames are marked with a solid circle (i.e. ). As can be seen, the heating time decreased exponentially with an increase in thermal input power. The correlation between thermal input power and heating time was not linear, but rather was logarithmic. This is thought to be related to the variation of specific heat or heat capacity of burned gas because the specific heat capacity of species i (cp,i) is inversely proportional to the temperature of species i (i.e. cp,i ~ 1/Ti). Fig. 10(a) shows the pollutant emission trends from an oxygenenhanced combustor. The pollutant level of NO was normalized using the emission index of NO (EINO). The EINO increased with an increase in fG ¼ 0.72 to 0.95 while it decreased with an increase in fG ¼ 0.95 to 1.03. This trend is thought to be reasonable in general because thermal NO is dominant in high temperature conditions (over 1300 C) and burned gas temperature increases in near stoichiometric conditions (approximately fG ¼ 1.0). EINO was emitted in oxy-fuel conditions which is marked as a solid circle (i.e. ) in Fig. 9(a) even though the emission level was lower than that in oxygen-enhanced conditions. This is thought to be due to the oxidation of a small amount of N2 in NG (i.e. 0.1% N2) in pure oxygen conditions. Even a small amount of N2 addition in an oxy-fuel flame can lead to emit high level of nitrogen oxides because the adiabatic temperature of an oxy-CH4 flame is higher about 800 C than that of a CH4-air flame [21]. Fig. 10(b) shows the combustion efficiency trend in an oxygenenhanced combustor. The equation for combustion efficiency (3 c) was defined as follows [6];
3c
¼
XCO2 at 15% O2 100% XCO2 at 15% O2 þ XCO at 15% O2
(7)
where Xi at 15% O2 is the mole fraction of species i normalized at XO2 ¼ 15%: The Xi at 15% O2 was defined as follow;
Fig. 9. Performance curve of the pilot-scale oxygen-enhanced combustor when varying the furnace heating time to Ttot ¼ 1250 C in Cases 3e13.
J. Oh et al. / Applied Thermal Engineering 91 (2015) 749e758
757
5. Conclusion In the current study, the performance of a pilot-scale oxygenenhanced furnace for scrap preheating was numerically and experimentally investigated with a multi-nozzle inverse diffusion burner. NG, composed of 91.6% CH4, 5.8% C2H6, 1.7% C3H8, 0.8% C4H10, and 0.1% N2, was used as a fuel. Oxygen was diluted with air in the rage of XO2 ¼ 21%e100%: Thermal input power varied in QTh ¼ 120e300 Mcal/h. The conclusions are summarized as follows: (1) Hot burned gas injected from an oxygen-enhanced burner was well-stirred in a furnace and heat-up the scrap preheat zone homogeneously. The volume-averaged temperature in the furnace was predicted at about 1250 C in steady state conditions. (2) The tendency of temperature rising in the furnace increased steeply with an increase in the oxygen mole fraction in an oxidant. The characteristic time for furnace heat-up increased exponentially as the thermal input power increased. The correlation between thermal input power and heating time was logarithmic, not linear. (3) The EINO increased with an increase in fG ¼ 0.72 to 0.95 while EINO decreased with an increase in fG ¼ 0.95 to 1.03. Even when a small amount of N2 was added to an oxy-fuel flame, this led to a high level of nitrogen oxides in comparison with an oxygen-enhanced flame. Acknowledgements This work was supported by an Energy Efficiency & Resources Core Technology Program of the Korea Institute of Energy Technology Evaluation and Planning (KETEP) grant funded by the Korean Government Ministry of Knowledge Economy (NP20130026-3). References
Fig. 10. Performance of a pilot-scale oxygen-enhanced combustor; (a) pollutant emission level and (b) combustion efficiency under the varying operating conditions in Table 3.
Xi at 15% O2 ¼ Xi
XO2 ;in 15% 100% XO2 ;in XO2 ;out
(8)
where Xi is the mole fraction of species i, XO2 ;in and XO2 ;out are the mole fraction of O2 in an oxidant and flue gas, respectively. The combustion efficiency of non-premixed oxygen-enhanced flames was maximized as 99.4% in fG ¼ 0.72 for both cases of oxyNG combustion and OEC. Meanwhile, the combustion efficiency in other cases was 62.7e81.5 in fG ¼ 0.50, 0.95, and 1.03. An oxygenenhanced flame in fG ¼ 0.95 and 1.03 is considered to be locally incomplete combustion even though the global equivalence is near stoichiometric conditions. In the case of fG ¼ 0.50, the oxygenenhanced flame is thought to be near blow-out region because EINO was maximized even in fuel-lean conditions. This means that a locally fuel-rich flame base was stabilized near the nozzle exit while local flame extinction occurred around the flame tip region due to the highly stretched flow.
[1] L. Zheng, Oxy-fuel Combustion for Power Generation and Carbon Dioxide (CO2) Capture, Woodhead Publishing Limited, Cambridge, UK, 2011. [2] Y.N. Toulouevski, I.Y. Zinurov, Innovation in Electric Arc Furnaces, SpringerVerlag, Berlin, Germany, 2010. [3] C.E. Baukal, Oxygen-enhanced Combustion, CRC Press, Boca Raton, USA, 2000. [4] Website address: https://steelplantech.com/product/ecoarc/. [5] Website address: http://www.corefurnace.com/meltshop_01.html. [6] K.K. Wu, Y.C. Chang, C.H. Chen, Y.D. Chen, High-efficiency combustion of natural gas with 21e30% oxygen-enriched air, Fuel 89 (2010) 2455e2462. k, Experimental study on the influence of [7] P. Bĕlohradský, P. Skryja, I. Huda oxygen content in the combustion air on the combustion characteristics, Energy 75 (2014) 116e126. [8] Y. Tan, M.A. Douglas, K.V. Thambimuthu, CO2 capturing using oxygen enhanced combustion strategies for natural gas power plants, Fuel 81 (2002) 1007e1016. [9] Y. Tu, H. Liu, S. Chen, Z. Liu, H. Zhao, C. Zheng, Effects of furnace chamber shape on the MILD combustion of natural gas, Appl. Therm. Eng. 76 (2015) 64e75. nchez, F. Cadavid, A. Amell, Experimental evaluation of a 20 kW oxygen [10] M. Sa enhanced self-regenerative burner operated in flameless combustion mode, Appl. Energy 111 (2013) 240e246. [11] S. Turns, An Introduction to Combustion: Concepts and Applications, third ed., McGraw-Hill, New York, USA, 2012, pp. 559e560. [12] J. Oh, S. Dong, J. Yang, The effect of fuel composition on the characteristics of a non-premixed synthetic natural gas-air flame, Fuel Process. Technol. 126 (2014) 66e75. [13] ANSYS FLUENT 12.0 Theory Guide, ANSYS Inc, Canonsburg, USA, 2009. [14] T.H. Shih, W.W. Liou, A. Shabbir, Z. Yang, J. Zhu, A new ke3 eddy-viscosity model for high Reynolds number turbulent flows - model development and validation, Comput. Fluids 24 (3) (1995) 227e238. [15] F.M. White, Fluid Mechanics, seventh ed., vol. 362e364, McGraw Hill, New York, NY, USA, 2011, pp. 471e473. [16] J.I. Seo, Y.T. Guahk, S.H. Bae, J.G. Hong, U.D. Lee, H.D. Shin, Experimental study on the oxygen combustion characteristics with CO2 feeding, J. Korean Soc. Combust. 20 (2) (2005) 26e34. [17] Website address: http://www.gaseq.co.uk/.
758
J. Oh et al. / Applied Thermal Engineering 91 (2015) 749e758
[18] D.J. Hautman, F.L. Dryer, K.P. Schug, I. Glassman, A multiple-step over all kinetic mechanism for the oxidation of hydrocarbons, Combust. Sci. Technol. 25 (1981) 219e235. [19] J. Oh, D. Noh, Flame characteristics of a non-premixed oxy-fuel jet in a labscale furnace, Energy 81 (2015) 328e343.
[20] A. Abbassi, Kh. Khoshmanesh, Numerical simulation and experimental analysis of an industrial glass melting furnace, Appl. Therm. Eng. 28 (2008) 450e459. [21] J. Oh, D. Noh, Laminar burning velocity of oxy-methane flames in atmospheric condition, Energy 45 (2012) 669e675.