- Email: [email protected]
CFD modeling of carbon combustion and electrode radiation in an electric arc furnace
Applied Thermal Engineering 90 (2015) 831e837
Contents lists available at ScienceDirect
Applied Thermal Engineering journal homepage: www.elsevier.com/locate/apthermeng
Research paper
CFD modeling of carbon combustion and electrode radiation in an electric arc furnace Cemil Yigit a, Gokhan Coskun a, *, Ekrem Buyukkaya a, Ufuk Durmaz a, Hasan R. Güven b a b
Department of Mechanical Engineering, Sakarya University, Sakarya 54187, Turkey _ _ Department of Mechanical Engineering, Istanbul University, Istanbul 54452, Turkey
h i g h l i g h t s 3D-CFD of carbon combustion and electrode radiation were performed for an EAF. Particles surface and gas phase reactions used to predict coal particle combustion. Realistic temperature distribution obtained for slag surface. Temperature, CO and CO2 distribution of carbon combustion visually obtained.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 2 March 2015 Accepted 23 July 2015 Available online 4 August 2015
This paper, concentrates on a three-dimensional (3D) computational fluid-dynamics (CFD) model for coal combustion and electrode radiation inside an electric-arc furnace (EAF). Simulation of the melting process includes combustion reactions of coal particles and radiation from electrodes. Particle surface and gas phase reactions were used to predict injected coal particle combustion. The predicted temperature distributions are realistic because of the inclusion of combustion reactions and radiation models together. The CFD model provided detail information for the coal particles combustion and radiation interactions phoneme inside the electric-arc furnace. The cooling process of the EAF walls were considered in the CFD model hence heat losses from the walls have been found higher than the earlier studies. Results showed that CFD simulation can efficiently be used to develop and investigate EAF in design phase. © 2015 Elsevier Ltd. All rights reserved.
Keywords: Electric arc furnace CFD Combustion Radiation Heat transfer
1. Introduction Today, energy is an expensive and valuable resource for the heavy industry operations. Approaching to the maximum efficiency is one of the critical targets for the industrial sustainability. Generally electric arc furnaces consume high levels of energy to run the steelmaking processes that uses the coal and the electricity as the main resources of the energy. In order to increase the efficiency of these types of huge processes expensive experimental investigations using trial-error methods are required. Recently developed computational resources can solve interpenetrating physical phenomenon such as the solid particles combustion and the evaporation as well as the turbulence and the radiation together. CFD is an advanced simulation tool with a capability to
* Corresponding author. E-mail address: [email protected] (G. Coskun). http://dx.doi.org/10.1016/j.applthermaleng.2015.07.066 1359-4311/© 2015 Elsevier Ltd. All rights reserved.
comprehensive model for the industrial applications such as EAF. In this study a commercial CFD software ANSYS FLUENT 14.0 [1] is used to develop a simulation methodology on EAF. One of the important targets to create an extensive CFD model of EAF is to investigate the combustion phoneme of injected coal particles, electric arc radiation and slag surface temperature distribution in order to analyze and verify the accuracy of simulations. In addition, the CFD model provided detailed information on the spatial distribution of the jet velocity, combustion species distributions and heat output from the EAF walls. A number of reference studies have been carried out on combustion for the EAF. A comprehensive CFD model on post combustion reactions in the EAF has been studied by Li and Fruehan [2]. The authors simulated the radiation heat transfer, post combustion reactions, and de-post combustion reaction which is reacted between CO2 and carbon in the liquid metal or the electrodes. In the light of this information, coal combustion reactions have been added and solved with using three gas phases and three char
832
C. Yigit et al. / Applied Thermal Engineering 90 (2015) 831e837
oxidation reactions in the current work. Radiation heat transfer from the electric-arc has significant effect to melt the metal mixture in the EAF. There are several reference studies to compute the radiation on heat transfer using CFD. Guo and Irons [3,4] simulated the EAF radiation using CFD in order to introduce the energy distribution of the radiation heat transfer. They studied the radiation energy distribution for the sidewall refractory, watercooled side panels and the furnace roof. Based on Guo's study some assumptions on heat transfer coefficients, radiation emissivity of the walls and the electric-arc surfaces were applied in the current model. Electric arc furnaces are widely used to melt scrap metal in the steelmaking industry. Taking this into account, the studies on energy efficiency are of a great significance. Therefore, this study aims to develop a comprehensive three dimensional CFD model for the coal combustion and the radiation heat transfer process. In this model, the location or quantity of the injectors, the injection angle, chamber dimensions of the EAF, the oxygen or the coal contents and radiation rates of the electrodes etc. can vary to reach the targeted temperature on the melt surface. On that sense the model provides flexibility on design parameters. Such a flexible CFD model can provide optimization the combustion and the electrode radiation to decrease the energy consumption in an EAF. In this study, an EAF produced by CVS Technologies for scrap melting process has been modeled.
Table 1 EAF parameters. Internal height (mm) Internal furnace radius (mm) Internal slag surface radius (mm) Electrode radius (mm) Coal injector diameter (mm) Oxygen injector diameter inner e outer (mm) Injector angle e Three Oxygen flow rate (kg/s/injector) Coal flow rate (kg/s/injector)
3505.25 3600 3080 305 25.4 68.14e45.4 45 downwards 1.2955 0.333
2. CFD model 2.1. Geometry The EAF parameters which are given in Table 1 were used to create the 3-D model for the CFD simulation. Boundary conditions and angles on x, y coordinate of injectors were given particularly in Fig. 1. Injector 1, 2 and 3 were positioned on þ X axis 25 , 25 , 102 respectively. All injectors located 1.07 m above from the slag surface and aiming 45 downwards. Coal particles and O2 were sent into the furnace through the injectors as shown in bottom-right of Fig. 1. One of the main purposes of the model is to obtain upper surface temperature distribution of the melt layer. In order to simplify the model, bottom surface of the computational domain was accepted as a slag surface instead of modeling the overall melt volume. Computational domain borders also include inner surfaces of the furnace walls, upper and side walls of the chimney. Combustion products and the other unburned gasses leave the computational domain from the exhaust surface of the EAF. Bottom surfaces of three electrodes generate electric arc and the radiate energy with the constant heat flux.
2.2. Computational mesh Reliability of the CFD analysis results is directly related to quality of the mesh structure. The skewness is an important parameter to understand the grid structure quality therefore great attention has been given to the skewness. Triangle prism mesh structures were used to create small computational domains in the model. The boundary layer mesh and weight factors are also used to create some critical zones for the computational accuracy. The sufficient grid refinement at the slag surface, arc generated surfaces of the electrodes and the injectors are necessary to capture the surface interactions and provide computational stability. The mesh structure of the EAF model, mesh inflation at the slag surface and dense grids at the injectors can be seen in Fig. 2. Sensitivity of different grid densities, with the total number of cells ranging about 560,000e1,130,000 is tested for the same
Exhaus Electrodes Furnace Wall
Injector 1 Injector 2
Upper Bath Injector 3
Oxygen
Lower Bath Slag surface Fig. 1. 3-D model of the EAF.
Carbon
C. Yigit et al. / Applied Thermal Engineering 90 (2015) 831e837
833
Fig. 2. EAF mesh structure (left). Dense mesh structures for injector area (right-up) and boundary layer approach above the slag surface (right-down).
model. The results indicated that a sufficiently fine grid is achieved with 733,000 cells. The average temperature on the slag surface no longer changed by increasing the grid size as indicated in Table 2.
reaction [5,6] into account, CO applied as 0.021455 kg/m3s as an inlet for the slag surface. 2.4. Numerical model of the EAF
2.3. Boundary conditions The standard wall function was applied to each wall boundaries. 873 K and 773 K static temperatures applied for the lower and the upper bath walls respectively. Considering the cooling process, static temperature was applied as 393 K to the furnace wall. 1.15e þ 08 W/m2 heat flux was applied at the bottom surface of the electrodes in order to model 100 MW electricity. To obtain morerealistic thermal boundary conditions convection to the outer ambient and conduction on the melt layer were considered in the model by activation convection and conduction heat transfer equations at the slag surface boundary. Convection heat transfer coefficient was used as 30 W/m2K, thermal conductivity was used as 0.5 W/mK. The radiation internal emissivity was applied to the each wall in the model. The internal emissivity as the boundary condition was used as 0.7 for the slag surface, the lower and the upper bath walls and for the granite electrode walls was 0.85 [3]. As shown in Figs. 1 and 2, the outer tube feeds the oxygen enriched air and defines as velocity inlet boundary condition. A constant velocity of the oxygen enriched air is 480 m/s and oxygen mass fraction is 0.3871. Initial temperature of the air was set as 1500 K to activate the combustion reactions. Inner tube was defined as an injection model which feeds the coal particles. Velocity and temperature of the coal particles are 0.7591 m/s and 300 K respectively. The CO gas emerges from the slag surface to inside the furnace therefore the slag surface is treated as an inlet boundary condition. In a typical heat in the EAF, about 1.5 kmol of CO is generated for per ton of steel [2]. Taking CO produced from FeO
Table 2 Average results from slag surface of different grid densities. Number of mesh Radiation Static elements temperature (K) temperature (K)
Mass fraction C of mvvol concentration
564,657 733,962 1,129,940
0.14124 0.119 0.1186
2045 2045 2048
1880 1882 1884
0.00026661 0.0002215 0.000216
The model of EAF includes different types of mathematical methods as the turbulent flows, combustion reactions and the radiation heat transfer, etc. Solving all these equations from the beginning of the solution for the initial conditions simultaneously is almost impossible for converging. In order to resolve this problem equations were activated progressively. In the beginning, flow analysis run for the sufficient iterations, later combustion equations and finally radiation equations were activated. The pressure velocity coupling is achieved through the COUPLED e algorithm. The realizable keε model was applied in the model to solve the turbulence which shows a good agreement with experimental observations of the gas flows with the coal particles combustion according to literature [7]. Previous studies on the liquid fuels and the coal combustion modeling [8e10] were showed that discrete phase model (DPM) which has applied in current study, is a useful method to simulate the particle trajectory. While the coal particle phase is treated with an Eulerian description needs lower CPU then Lagrangian approach without any considerable differences [11]. Therefore Eulerian approach was considered in the current model. The particle diameter distribution of the coal was modeled using Rosin-Rammer method with spread parameter of 4.52 and 10 groups. The minimum, the maximum and the mean particle diameters were chosen as respectively 70e-06 m, 200e-06 m and 134e-06 m. Discrete random walk model was enabled for stochastic tracking of coal particles. From the literature on CFD simulation of the EAF, Scheepers et al. [12] showed P1 radiation model in the Fluent is able to solve the radiation effectively. Thus the P1 radiation model has been chosen in order to solve the radiation originated from the electrodes and combustion reactions in the current model. The conservation of the mass is solved for each relevant chemical species. Reaction rates are calculated using the Finite-Rate Eddy-Dissipation model [13] considering that the rate-controlling mechanism which may be either the chemical kinetics or the turbulent mixing. In this way, the effect that local micro-mixing may have in the determination of the rate of reaction [14]. In the CFD code, the coal is simply classified as a single species due to the quantities of the fixed carbon and volatile matter [1].
834
C. Yigit et al. / Applied Thermal Engineering 90 (2015) 831e837
Table 3 Gas phase reactions. Number
Reaction
A
Ea (j/kgmol)
1 2 3
mvvol þ 1:706O2 /CO2 þ 1:543H2 O H2 þ 0; 5O2 /H2 O CO þ 0; 5O2 /CO2
2.119eþ11 1eþ15
2.027eþ08 1eþ08
Several classifications for the coal can be found in the FLUENT depending on property of the coal density, the specific heat and component fractions etc. In this study the volatile matter (VM) break up was assumed as a single hypothetic hydrocarbon component consist of Carbon (C), Hydrogen (H), and Oxygen (O). After injecting the coal particles into the computational volume, volatile matters in the coal particles are initially converted to a pseudo gas phase species with using the constant rate devolatilization model [15]. Three-step volumetric reactions with six species including the volatiles matter combustion were applied to the model. These gas phase reactions deals with the local chemical equilibrium are calculated by Arrhenius type chemical kinetics. Three-step gas phase reactions were listed in Table 3 including Arrhenius Reaction Rate and Activation Energy. When component of the volatiles matter is completely spread out from the particle surfaces, solid phase reactions begin [16]. Assuming all solid matter composition is carbon in the model, complex combustion mechanism of the char can be represented by a simple three heterogeneous reactions to reduce the computational time. Surface reactions, as listed in Table 4, are solved in the CFD code by using first-order global Arrhenius kinetics [7] and the multiple surface reactions model [1]. Sum of the carbon consumption rates from each individual surface reaction equals to the total carbon consumption rate at the computational volume [17]. 3. Results and discussion 3.1. The flow field inside the EAF Fig. 3(a) shows the velocity profiles on a plane intersecting the center of the injector 1. Velocity data plotted in the Fig. 3(b) is obtained from a straight line (demonstrated in the Fig. 3(a) with a dash line) from the slag surface to the injector. Values range from a maximum from 349 m/s to zero for the O2 stream and outer flow. The speed of the jet decreases as it expands and mixes with the coal particles and gases. Swirls and high temperature combustion effects expand the velocity gradients to the larger volumes. Jet reaches the melt surface about 75 m/s. 3.2. The thermal field inside the EAF The temperature profiles inside the EAF are presented for a plane intersecting from the center of the injectors. The combustion temperature profiles of injector 1, 2 and 3 are shown in Fig. 4. The temperature at the jet flow after injector 1 and 3 is distributed between 2270 and 3200 K. Injector 2 which is located between two
Table 4 Char oxidation model of the coal particles.
4 5 6
A
3.3. Mass fraction of species inside the EAF The gas and the solid particle concentrations in the EAF along a line which is created from the slag surface (0 m) to the center of injector 1 exit (1.2 m) are given in Fig. 8. Some of the solid coal particles are evaporated rapidly as soon as they leave the injectors while the surface reactions are started by the rest of them. The surface reactions occur at 0.5 m away from the injector's exit because of injector's design. Rapidly evaporated coal particles react with the oxygen at high temperatures. As a result of the combustion reactions, the mass fraction of the carbon dioxide increases to 0.35 while most of the O2 was consumed. The mass fraction of the evaporated coal particles (the gas phase coal) dramatically decreases about 0.3 m away from the injector due to high temperature combustion. The carbon monoxide is one of the most important species for the combustion reactions in the EAF. CO consists of two different sources which are combustion of the coal particles and diffusion from the slag surface. Some of the CO composed by the gas phase and char reactions of the coal and CO2 species. The other source is CO diffusion of the foamy slag coal layer at the slag surface [6]. CO diffusion is originated from the injected coal and O2 among to the metal and the slag interface. Fig. 9 shows CO and CO2 mass fraction distribution around the injector 1. Formation of CO from combustion reactions is lower than the slag surface diffusion as indicated in Fig. 9 (left). Additionally the beginning of the decline in the mass fraction of CO2 right after high temperature region showing the char oxidation (reaction 5 in Table 4) build up in the accumulated coal particles above the slag surface. 3.4. Validation
Heterogeneous particle surface reactions Number Reaction
injectors, the maximum combustion temperature at the jet flow is investigated as 3000 K. The minimum temperature values are obtained at the near furnace walls as 692 K because of the heat transfer effects from the cooled furnace walls. The maximum combustion temperatures at the injector exits are continuously decreased to around 2200 K on the slag surface. The radiation temperature distribution which is identified on the bottom surface of the electrode can be seen in Fig. 5. The radiation temperature range is around 150 K inside the furnace chamber. Probability Density Function (PDF) and the temperature distribution on the slag surface were introduced in Fig. 6. According to PDF figure the slag surface static temperature variation is between 1860 K and 1970 K and the temperature of the slag surface mostly accumulated around 1880 K which is enough to maintain melt temperature at around 1800 K [18]. Fig. 7 shows the heat transfer distribution by the radiation and the convection to the EAF walls, the slag surface and the chimney. The heat generated by the radiation and the combustion are 100.75 MW and 5 MW, respectively. The heat transferred by the radiation to the EAF's wall and the roof are 65,190 and 26,068 MW respectively. The figure also shows that 8964 MW thermal energy is transferred by the radiation to the metal slag surface. Besides that the heat loss from the chimney via off gas is only 2348 MW. The heat loss from the convection was very low in percentage compared with the radiation heat transfer therefore a detailed diagram that depicts the convection heat transfer is given in Fig. 7 (right).
Ea (j/kgmol) Diffusion rate constant
0.002 7.9eþ07 C < s > þ 0; 5O2 /CO C < s > þ CO2 /2CO C < s > þ H2 O/H2 þ CO
5e-12
In the current study the coal combustion was simulated by using same reactions with Zhang et al. [7] which was validated with their experimental results. To compare of these two studies proportional gas temperature differences were obtained from belong figures. Totally five different data points were selected. The highest
C. Yigit et al. / Applied Thermal Engineering 90 (2015) 831e837
835
Fig. 3. Velocity (m/s) magnitude for a plane intersecting the injector 1; (a) Velocity distribution profile and (b) Velocity data graph.
Fig. 4. Temperature (K) profile for a plane intersecting the injector 1, 2 and 3.
Fig. 5. Radiation temperature (K) profile for a plane intersecting one of the electrodes.
temperature in the core of the jet marked as point 3 of the injector 1 (Fig. 4) is 3200 K. The nearest ambient temperature to the jet flame marked as point 1 is about 1960 K. The average temperature of jet flame marked as point 5 is about 2580 K. Points 2 and 4 are geometrically middle points. Each data points were divided by the maximum temperature value in order to obtain proportional gas temperature differences. Same procedure was applied for Figure 17 in Ref. [7]. The graphical comparison consisting of proportional gas temperature differences from the simulation results was given in
Fig. 10. Simulation results in the temperature field are compatible with each other as understand from the figure. Li and Fruehan obtained distribution of the heat output from the CFD simulation and they gave the graphical result in Ref. [2]. They obtained 12% of the total energy transferred to the metal by the radiation while in current study is about % 9. This situation shows that just about 10% of the total energy can be transferred to the metal. On the other hand reference [2] obtained 45% of the total energy transferred to the wall by the radiation while in current
836
C. Yigit et al. / Applied Thermal Engineering 90 (2015) 831e837
Fig. 6. Static temperature (K) and PDF distribution profile on the slag surface.
Fig. 7. Heat output obtained from convection, radiation and off-gas.
study is about 65%. This difference is due to applying lower and varying thermal conditions on the furnace walls. Already applying different wall temperatures to the EAF walls for more-realistic thermal boundary conditions was aforementioned in Ref. [2]. Also Istadi and Bindar [19] focused on cooler design of an EAF using CFD modeling. They obtained cooled furnace wall temperature distribution around 423 K which is almost equal to the current model's boundary condition. Furthermore geometry of the EAF and combustion model differences also caused energy distribution variations on other boundaries between current and reference models [2].
4. Conclusion
Fig. 8. Coal gas, coal particles, carbon monoxide, carbon dioxide and oxygen at the center of the injector 1 jet.
Experimental analysis of factors such as the combustion, the chemical reactions and the temperature distribution are effected by the radiation in a complicated manner. This is primary importance on the efficiency of an Electric Arc Furnace. This is also extraordinary expensive. Detailed examination of the processes that occurred in the existing furnace is very significant to develop and become more efficient.
C. Yigit et al. / Applied Thermal Engineering 90 (2015) 831e837
837
Fig. 9. CO (left) and CO2 (right) mass fraction in combustion field on a plane intersection the injector 1.
Acknowledgements
Proportional Gas Temperature Differences
1.2 Model 1
This work was supported by the CVS Technologies at GebzeKocaeli, TURKEY. We would like to thank the CVS for their support.
Zhang et al. [7]
0.8 0.6
References
0.4
[1] ANSYS, Inc, FLUENT, Version 14.0, ANSYS, Inc, Canonsburg, PA, 2013. [2] Y. Li, R. Fruehan, Computational fluid dynamics simulation of post combustion in the electric arc furnace, Metallurgical Mater. Trans. 34 (3) (2003) 333e343. [3] D. Guo, G. Irons, Modeling of radiation intensity an EAF, in: Third International Conference on CFD in the Minerals and Process Industries, Melbourne, Australia, 2003, pp. 10e12. €nsson, Modeling of a DC electric arc [4] J. Alexis, M. Ramirez, G. Trapaga, P. Jo furnace - heat transfer from the arc, ISIJ Int. 40 (2000) 1089e1097. [5] M. Rahman, Fundamental Investigation of Slag/Carbon Interactions in Electric Arc Furnace Steelmaking Process (PhD Thesis), The University of New South Wales, 2010. [6] M. Grant, Principles and strategy of EAF post combustion, in: Electric Furnace Conference, Orlando, USA, 2000, pp. 12e15. [7] J. Zhang, W. Prationo, L. Zhang, Z. Zhang, Computational fluid dynamics modeling on the air-firing and oxy-fuel combustion of dried victorian brown coal, Energy Fuels 27 (2013) 4258e4269. jek, Validation of an effervescent spray model with secondary [8] J. Broukal, J. Ha atomization and its application to modeling of A large-scale furnace, Appl. Therm. Eng. 31 (2011) 2153e2164. [9] R.I. Singh, A. Brink, M. Hupa, CFD modeling to study fluidized bed combustion and gasification, Appl. Therm. Eng. 52 (2) (2013) 585e614. [10] B. Rahmanian, M.R. Safaei, S.N. Kazi, G. Ahmadi, H.F. Oztop, K. Vafai, Investigation of pollutant reduction by simulation of turbulent non-premixed pulverized coal combustion, Appl. Therm. Eng. 73 (1) (2014) 1222e1235. [11] A.C. Benim, B. Epple, B. Krohmer, Modelling df pulverised coal combustion by a eulerian-eulerian two-phase flow formulation, Prog. Comput. Fluid Dyn. 5 (6) (2005) 345e361. [12] E. Scheepers, Y. Yang, A.T. Adema, R. Boom, M.A. Reuter, Process modeling and optimization of a submerged arc furnace for phosphorus production, Metallurgical Mater. Trans. B 41 (B) (2010) 990e1005. [13] B. Magnussen, B. Hjertager, On the structure of turbulence and A generalised eddy dissipation concept for chemical reaction in turbulent flow, in: AIAA Aerospace Meeting, St. Louis, USA, 1981, pp. 12e15. [14] D.L. Marchisio, A.A. Barresi, Understanding coal gasification and combustion modeling in general purpose CFD code, IEEE Trans. Plasma Sci. 58 (2003) 3579e3587. [15] M.M. Baum, P.J. Street, Predicting the combustion behavior of coal particles, Combust. Sci. Technol. 3 (5) (1971) 231e243. [16] L. Hookyung, S. Choi, B.J. Kim, Coal gasification and combustion modeling in general purpose CFD code, J. Korean Soc. Combust. 15 (3) (2010) 15e24. [17] J. Zhang, M. Zhang, J. Yu, Extended application of the moving flame front model for combustion of a carbon particle with a finite-rate homogenous reaction, Energy Fuels 24 (2010) 871e879. [18] E. Gomez, D. Rani, C. Cheeseman, D. Deegan, M. Wise, A. Boccaccini, Thermal plasma technology for the treatment of wastes: a critical review, J. Hazard. Mater. 161 (2e3) (2009) 614e626. [19] I. Istadia, Y. Bindarb, Improved cooler design of electric arc furnace refractory in mining Industry using thermal analysis modeling and simulation, Appl. Therm. Eng. 73 (1) (2014) 1129e1140.
0.2 0 0
1
2
3
4
5
6
Data Point Fig. 10. Proportional gas temperature differences for each data point. Comparison of the current simulation and Zahng et al. [7].
In this study, the thermal effects of the coal combustion and the electrode based arc radiation are modeled for a commercial EAF using the state of art simulation technique of CFD. Core temperature occurred with the combustion in the jet and the temperature distribution on the slag surface has been analyzed in detail. The variation of the CO and CO2 which are occurred by the combustion reactions and diffusion from the slag surface are given numerically and graphically. The effect of the radiation on the temperature distribution inside the EAF and on the slag surface is examined. The total and the average temperature distribution on the slag surface is around 1770 K. It is important to keep scrap metal at the fluid phase for the better quality. The average temperature at the slag surface is 1880 K while the minimum temperature is 1770 K. This temperature distribution meets the requirements. The developed model can help us to increase the thermal efficiency of an EAF by testing at the design level. It was determined that the model is fast, reliable, low cost and able to give detailed results. This model gives us an economic alternative against the costly experimental studies and also gives realistic and detailed results against the zero dimensional calculations which have the less consistency. It is seen that the major heat transfer mechanism is the radiation predictably. As a result of the study the heat transferred to the metal slag surface is very poor in comparison with the heat transferred to the EAF walls and the roof. The radiation heat loss from the EAF's wall and the roof is considerable.
Contents lists available at ScienceDirect
Applied Thermal Engineering journal homepage: www.elsevier.com/locate/apthermeng
Research paper
CFD modeling of carbon combustion and electrode radiation in an electric arc furnace Cemil Yigit a, Gokhan Coskun a, *, Ekrem Buyukkaya a, Ufuk Durmaz a, Hasan R. Güven b a b
Department of Mechanical Engineering, Sakarya University, Sakarya 54187, Turkey _ _ Department of Mechanical Engineering, Istanbul University, Istanbul 54452, Turkey
h i g h l i g h t s 3D-CFD of carbon combustion and electrode radiation were performed for an EAF. Particles surface and gas phase reactions used to predict coal particle combustion. Realistic temperature distribution obtained for slag surface. Temperature, CO and CO2 distribution of carbon combustion visually obtained.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 2 March 2015 Accepted 23 July 2015 Available online 4 August 2015
This paper, concentrates on a three-dimensional (3D) computational fluid-dynamics (CFD) model for coal combustion and electrode radiation inside an electric-arc furnace (EAF). Simulation of the melting process includes combustion reactions of coal particles and radiation from electrodes. Particle surface and gas phase reactions were used to predict injected coal particle combustion. The predicted temperature distributions are realistic because of the inclusion of combustion reactions and radiation models together. The CFD model provided detail information for the coal particles combustion and radiation interactions phoneme inside the electric-arc furnace. The cooling process of the EAF walls were considered in the CFD model hence heat losses from the walls have been found higher than the earlier studies. Results showed that CFD simulation can efficiently be used to develop and investigate EAF in design phase. © 2015 Elsevier Ltd. All rights reserved.
Keywords: Electric arc furnace CFD Combustion Radiation Heat transfer
1. Introduction Today, energy is an expensive and valuable resource for the heavy industry operations. Approaching to the maximum efficiency is one of the critical targets for the industrial sustainability. Generally electric arc furnaces consume high levels of energy to run the steelmaking processes that uses the coal and the electricity as the main resources of the energy. In order to increase the efficiency of these types of huge processes expensive experimental investigations using trial-error methods are required. Recently developed computational resources can solve interpenetrating physical phenomenon such as the solid particles combustion and the evaporation as well as the turbulence and the radiation together. CFD is an advanced simulation tool with a capability to
* Corresponding author. E-mail address: [email protected] (G. Coskun). http://dx.doi.org/10.1016/j.applthermaleng.2015.07.066 1359-4311/© 2015 Elsevier Ltd. All rights reserved.
comprehensive model for the industrial applications such as EAF. In this study a commercial CFD software ANSYS FLUENT 14.0 [1] is used to develop a simulation methodology on EAF. One of the important targets to create an extensive CFD model of EAF is to investigate the combustion phoneme of injected coal particles, electric arc radiation and slag surface temperature distribution in order to analyze and verify the accuracy of simulations. In addition, the CFD model provided detailed information on the spatial distribution of the jet velocity, combustion species distributions and heat output from the EAF walls. A number of reference studies have been carried out on combustion for the EAF. A comprehensive CFD model on post combustion reactions in the EAF has been studied by Li and Fruehan [2]. The authors simulated the radiation heat transfer, post combustion reactions, and de-post combustion reaction which is reacted between CO2 and carbon in the liquid metal or the electrodes. In the light of this information, coal combustion reactions have been added and solved with using three gas phases and three char
832
C. Yigit et al. / Applied Thermal Engineering 90 (2015) 831e837
oxidation reactions in the current work. Radiation heat transfer from the electric-arc has significant effect to melt the metal mixture in the EAF. There are several reference studies to compute the radiation on heat transfer using CFD. Guo and Irons [3,4] simulated the EAF radiation using CFD in order to introduce the energy distribution of the radiation heat transfer. They studied the radiation energy distribution for the sidewall refractory, watercooled side panels and the furnace roof. Based on Guo's study some assumptions on heat transfer coefficients, radiation emissivity of the walls and the electric-arc surfaces were applied in the current model. Electric arc furnaces are widely used to melt scrap metal in the steelmaking industry. Taking this into account, the studies on energy efficiency are of a great significance. Therefore, this study aims to develop a comprehensive three dimensional CFD model for the coal combustion and the radiation heat transfer process. In this model, the location or quantity of the injectors, the injection angle, chamber dimensions of the EAF, the oxygen or the coal contents and radiation rates of the electrodes etc. can vary to reach the targeted temperature on the melt surface. On that sense the model provides flexibility on design parameters. Such a flexible CFD model can provide optimization the combustion and the electrode radiation to decrease the energy consumption in an EAF. In this study, an EAF produced by CVS Technologies for scrap melting process has been modeled.
Table 1 EAF parameters. Internal height (mm) Internal furnace radius (mm) Internal slag surface radius (mm) Electrode radius (mm) Coal injector diameter (mm) Oxygen injector diameter inner e outer (mm) Injector angle e Three Oxygen flow rate (kg/s/injector) Coal flow rate (kg/s/injector)
3505.25 3600 3080 305 25.4 68.14e45.4 45 downwards 1.2955 0.333
2. CFD model 2.1. Geometry The EAF parameters which are given in Table 1 were used to create the 3-D model for the CFD simulation. Boundary conditions and angles on x, y coordinate of injectors were given particularly in Fig. 1. Injector 1, 2 and 3 were positioned on þ X axis 25 , 25 , 102 respectively. All injectors located 1.07 m above from the slag surface and aiming 45 downwards. Coal particles and O2 were sent into the furnace through the injectors as shown in bottom-right of Fig. 1. One of the main purposes of the model is to obtain upper surface temperature distribution of the melt layer. In order to simplify the model, bottom surface of the computational domain was accepted as a slag surface instead of modeling the overall melt volume. Computational domain borders also include inner surfaces of the furnace walls, upper and side walls of the chimney. Combustion products and the other unburned gasses leave the computational domain from the exhaust surface of the EAF. Bottom surfaces of three electrodes generate electric arc and the radiate energy with the constant heat flux.
2.2. Computational mesh Reliability of the CFD analysis results is directly related to quality of the mesh structure. The skewness is an important parameter to understand the grid structure quality therefore great attention has been given to the skewness. Triangle prism mesh structures were used to create small computational domains in the model. The boundary layer mesh and weight factors are also used to create some critical zones for the computational accuracy. The sufficient grid refinement at the slag surface, arc generated surfaces of the electrodes and the injectors are necessary to capture the surface interactions and provide computational stability. The mesh structure of the EAF model, mesh inflation at the slag surface and dense grids at the injectors can be seen in Fig. 2. Sensitivity of different grid densities, with the total number of cells ranging about 560,000e1,130,000 is tested for the same
Exhaus Electrodes Furnace Wall
Injector 1 Injector 2
Upper Bath Injector 3
Oxygen
Lower Bath Slag surface Fig. 1. 3-D model of the EAF.
Carbon
C. Yigit et al. / Applied Thermal Engineering 90 (2015) 831e837
833
Fig. 2. EAF mesh structure (left). Dense mesh structures for injector area (right-up) and boundary layer approach above the slag surface (right-down).
model. The results indicated that a sufficiently fine grid is achieved with 733,000 cells. The average temperature on the slag surface no longer changed by increasing the grid size as indicated in Table 2.
reaction [5,6] into account, CO applied as 0.021455 kg/m3s as an inlet for the slag surface. 2.4. Numerical model of the EAF
2.3. Boundary conditions The standard wall function was applied to each wall boundaries. 873 K and 773 K static temperatures applied for the lower and the upper bath walls respectively. Considering the cooling process, static temperature was applied as 393 K to the furnace wall. 1.15e þ 08 W/m2 heat flux was applied at the bottom surface of the electrodes in order to model 100 MW electricity. To obtain morerealistic thermal boundary conditions convection to the outer ambient and conduction on the melt layer were considered in the model by activation convection and conduction heat transfer equations at the slag surface boundary. Convection heat transfer coefficient was used as 30 W/m2K, thermal conductivity was used as 0.5 W/mK. The radiation internal emissivity was applied to the each wall in the model. The internal emissivity as the boundary condition was used as 0.7 for the slag surface, the lower and the upper bath walls and for the granite electrode walls was 0.85 [3]. As shown in Figs. 1 and 2, the outer tube feeds the oxygen enriched air and defines as velocity inlet boundary condition. A constant velocity of the oxygen enriched air is 480 m/s and oxygen mass fraction is 0.3871. Initial temperature of the air was set as 1500 K to activate the combustion reactions. Inner tube was defined as an injection model which feeds the coal particles. Velocity and temperature of the coal particles are 0.7591 m/s and 300 K respectively. The CO gas emerges from the slag surface to inside the furnace therefore the slag surface is treated as an inlet boundary condition. In a typical heat in the EAF, about 1.5 kmol of CO is generated for per ton of steel [2]. Taking CO produced from FeO
Table 2 Average results from slag surface of different grid densities. Number of mesh Radiation Static elements temperature (K) temperature (K)
Mass fraction C
564,657 733,962 1,129,940
0.14124 0.119 0.1186
2045 2045 2048
1880 1882 1884
0.00026661 0.0002215 0.000216
The model of EAF includes different types of mathematical methods as the turbulent flows, combustion reactions and the radiation heat transfer, etc. Solving all these equations from the beginning of the solution for the initial conditions simultaneously is almost impossible for converging. In order to resolve this problem equations were activated progressively. In the beginning, flow analysis run for the sufficient iterations, later combustion equations and finally radiation equations were activated. The pressure velocity coupling is achieved through the COUPLED e algorithm. The realizable keε model was applied in the model to solve the turbulence which shows a good agreement with experimental observations of the gas flows with the coal particles combustion according to literature [7]. Previous studies on the liquid fuels and the coal combustion modeling [8e10] were showed that discrete phase model (DPM) which has applied in current study, is a useful method to simulate the particle trajectory. While the coal particle phase is treated with an Eulerian description needs lower CPU then Lagrangian approach without any considerable differences [11]. Therefore Eulerian approach was considered in the current model. The particle diameter distribution of the coal was modeled using Rosin-Rammer method with spread parameter of 4.52 and 10 groups. The minimum, the maximum and the mean particle diameters were chosen as respectively 70e-06 m, 200e-06 m and 134e-06 m. Discrete random walk model was enabled for stochastic tracking of coal particles. From the literature on CFD simulation of the EAF, Scheepers et al. [12] showed P1 radiation model in the Fluent is able to solve the radiation effectively. Thus the P1 radiation model has been chosen in order to solve the radiation originated from the electrodes and combustion reactions in the current model. The conservation of the mass is solved for each relevant chemical species. Reaction rates are calculated using the Finite-Rate Eddy-Dissipation model [13] considering that the rate-controlling mechanism which may be either the chemical kinetics or the turbulent mixing. In this way, the effect that local micro-mixing may have in the determination of the rate of reaction [14]. In the CFD code, the coal is simply classified as a single species due to the quantities of the fixed carbon and volatile matter [1].
834
C. Yigit et al. / Applied Thermal Engineering 90 (2015) 831e837
Table 3 Gas phase reactions. Number
Reaction
A
Ea (j/kgmol)
1 2 3
mvvol þ 1:706O2 /CO2 þ 1:543H2 O H2 þ 0; 5O2 /H2 O CO þ 0; 5O2 /CO2
2.119eþ11 1eþ15
2.027eþ08 1eþ08
Several classifications for the coal can be found in the FLUENT depending on property of the coal density, the specific heat and component fractions etc. In this study the volatile matter (VM) break up was assumed as a single hypothetic hydrocarbon component consist of Carbon (C), Hydrogen (H), and Oxygen (O). After injecting the coal particles into the computational volume, volatile matters in the coal particles are initially converted to a pseudo gas phase species with using the constant rate devolatilization model [15]. Three-step volumetric reactions with six species including the volatiles matter combustion were applied to the model. These gas phase reactions deals with the local chemical equilibrium are calculated by Arrhenius type chemical kinetics. Three-step gas phase reactions were listed in Table 3 including Arrhenius Reaction Rate and Activation Energy. When component of the volatiles matter is completely spread out from the particle surfaces, solid phase reactions begin [16]. Assuming all solid matter composition is carbon in the model, complex combustion mechanism of the char can be represented by a simple three heterogeneous reactions to reduce the computational time. Surface reactions, as listed in Table 4, are solved in the CFD code by using first-order global Arrhenius kinetics [7] and the multiple surface reactions model [1]. Sum of the carbon consumption rates from each individual surface reaction equals to the total carbon consumption rate at the computational volume [17]. 3. Results and discussion 3.1. The flow field inside the EAF Fig. 3(a) shows the velocity profiles on a plane intersecting the center of the injector 1. Velocity data plotted in the Fig. 3(b) is obtained from a straight line (demonstrated in the Fig. 3(a) with a dash line) from the slag surface to the injector. Values range from a maximum from 349 m/s to zero for the O2 stream and outer flow. The speed of the jet decreases as it expands and mixes with the coal particles and gases. Swirls and high temperature combustion effects expand the velocity gradients to the larger volumes. Jet reaches the melt surface about 75 m/s. 3.2. The thermal field inside the EAF The temperature profiles inside the EAF are presented for a plane intersecting from the center of the injectors. The combustion temperature profiles of injector 1, 2 and 3 are shown in Fig. 4. The temperature at the jet flow after injector 1 and 3 is distributed between 2270 and 3200 K. Injector 2 which is located between two
Table 4 Char oxidation model of the coal particles.
4 5 6
A
3.3. Mass fraction of species inside the EAF The gas and the solid particle concentrations in the EAF along a line which is created from the slag surface (0 m) to the center of injector 1 exit (1.2 m) are given in Fig. 8. Some of the solid coal particles are evaporated rapidly as soon as they leave the injectors while the surface reactions are started by the rest of them. The surface reactions occur at 0.5 m away from the injector's exit because of injector's design. Rapidly evaporated coal particles react with the oxygen at high temperatures. As a result of the combustion reactions, the mass fraction of the carbon dioxide increases to 0.35 while most of the O2 was consumed. The mass fraction of the evaporated coal particles (the gas phase coal) dramatically decreases about 0.3 m away from the injector due to high temperature combustion. The carbon monoxide is one of the most important species for the combustion reactions in the EAF. CO consists of two different sources which are combustion of the coal particles and diffusion from the slag surface. Some of the CO composed by the gas phase and char reactions of the coal and CO2 species. The other source is CO diffusion of the foamy slag coal layer at the slag surface [6]. CO diffusion is originated from the injected coal and O2 among to the metal and the slag interface. Fig. 9 shows CO and CO2 mass fraction distribution around the injector 1. Formation of CO from combustion reactions is lower than the slag surface diffusion as indicated in Fig. 9 (left). Additionally the beginning of the decline in the mass fraction of CO2 right after high temperature region showing the char oxidation (reaction 5 in Table 4) build up in the accumulated coal particles above the slag surface. 3.4. Validation
Heterogeneous particle surface reactions Number Reaction
injectors, the maximum combustion temperature at the jet flow is investigated as 3000 K. The minimum temperature values are obtained at the near furnace walls as 692 K because of the heat transfer effects from the cooled furnace walls. The maximum combustion temperatures at the injector exits are continuously decreased to around 2200 K on the slag surface. The radiation temperature distribution which is identified on the bottom surface of the electrode can be seen in Fig. 5. The radiation temperature range is around 150 K inside the furnace chamber. Probability Density Function (PDF) and the temperature distribution on the slag surface were introduced in Fig. 6. According to PDF figure the slag surface static temperature variation is between 1860 K and 1970 K and the temperature of the slag surface mostly accumulated around 1880 K which is enough to maintain melt temperature at around 1800 K [18]. Fig. 7 shows the heat transfer distribution by the radiation and the convection to the EAF walls, the slag surface and the chimney. The heat generated by the radiation and the combustion are 100.75 MW and 5 MW, respectively. The heat transferred by the radiation to the EAF's wall and the roof are 65,190 and 26,068 MW respectively. The figure also shows that 8964 MW thermal energy is transferred by the radiation to the metal slag surface. Besides that the heat loss from the chimney via off gas is only 2348 MW. The heat loss from the convection was very low in percentage compared with the radiation heat transfer therefore a detailed diagram that depicts the convection heat transfer is given in Fig. 7 (right).
Ea (j/kgmol) Diffusion rate constant
0.002 7.9eþ07 C < s > þ 0; 5O2 /CO C < s > þ CO2 /2CO C < s > þ H2 O/H2 þ CO
5e-12
In the current study the coal combustion was simulated by using same reactions with Zhang et al. [7] which was validated with their experimental results. To compare of these two studies proportional gas temperature differences were obtained from belong figures. Totally five different data points were selected. The highest
C. Yigit et al. / Applied Thermal Engineering 90 (2015) 831e837
835
Fig. 3. Velocity (m/s) magnitude for a plane intersecting the injector 1; (a) Velocity distribution profile and (b) Velocity data graph.
Fig. 4. Temperature (K) profile for a plane intersecting the injector 1, 2 and 3.
Fig. 5. Radiation temperature (K) profile for a plane intersecting one of the electrodes.
temperature in the core of the jet marked as point 3 of the injector 1 (Fig. 4) is 3200 K. The nearest ambient temperature to the jet flame marked as point 1 is about 1960 K. The average temperature of jet flame marked as point 5 is about 2580 K. Points 2 and 4 are geometrically middle points. Each data points were divided by the maximum temperature value in order to obtain proportional gas temperature differences. Same procedure was applied for Figure 17 in Ref. [7]. The graphical comparison consisting of proportional gas temperature differences from the simulation results was given in
Fig. 10. Simulation results in the temperature field are compatible with each other as understand from the figure. Li and Fruehan obtained distribution of the heat output from the CFD simulation and they gave the graphical result in Ref. [2]. They obtained 12% of the total energy transferred to the metal by the radiation while in current study is about % 9. This situation shows that just about 10% of the total energy can be transferred to the metal. On the other hand reference [2] obtained 45% of the total energy transferred to the wall by the radiation while in current
836
C. Yigit et al. / Applied Thermal Engineering 90 (2015) 831e837
Fig. 6. Static temperature (K) and PDF distribution profile on the slag surface.
Fig. 7. Heat output obtained from convection, radiation and off-gas.
study is about 65%. This difference is due to applying lower and varying thermal conditions on the furnace walls. Already applying different wall temperatures to the EAF walls for more-realistic thermal boundary conditions was aforementioned in Ref. [2]. Also Istadi and Bindar [19] focused on cooler design of an EAF using CFD modeling. They obtained cooled furnace wall temperature distribution around 423 K which is almost equal to the current model's boundary condition. Furthermore geometry of the EAF and combustion model differences also caused energy distribution variations on other boundaries between current and reference models [2].
4. Conclusion
Fig. 8. Coal gas, coal particles, carbon monoxide, carbon dioxide and oxygen at the center of the injector 1 jet.
Experimental analysis of factors such as the combustion, the chemical reactions and the temperature distribution are effected by the radiation in a complicated manner. This is primary importance on the efficiency of an Electric Arc Furnace. This is also extraordinary expensive. Detailed examination of the processes that occurred in the existing furnace is very significant to develop and become more efficient.
C. Yigit et al. / Applied Thermal Engineering 90 (2015) 831e837
837
Fig. 9. CO (left) and CO2 (right) mass fraction in combustion field on a plane intersection the injector 1.
Acknowledgements
Proportional Gas Temperature Differences
1.2 Model 1
This work was supported by the CVS Technologies at GebzeKocaeli, TURKEY. We would like to thank the CVS for their support.
Zhang et al. [7]
0.8 0.6
References
0.4
[1] ANSYS, Inc, FLUENT, Version 14.0, ANSYS, Inc, Canonsburg, PA, 2013. [2] Y. Li, R. Fruehan, Computational fluid dynamics simulation of post combustion in the electric arc furnace, Metallurgical Mater. Trans. 34 (3) (2003) 333e343. [3] D. Guo, G. Irons, Modeling of radiation intensity an EAF, in: Third International Conference on CFD in the Minerals and Process Industries, Melbourne, Australia, 2003, pp. 10e12. €nsson, Modeling of a DC electric arc [4] J. Alexis, M. Ramirez, G. Trapaga, P. Jo furnace - heat transfer from the arc, ISIJ Int. 40 (2000) 1089e1097. [5] M. Rahman, Fundamental Investigation of Slag/Carbon Interactions in Electric Arc Furnace Steelmaking Process (PhD Thesis), The University of New South Wales, 2010. [6] M. Grant, Principles and strategy of EAF post combustion, in: Electric Furnace Conference, Orlando, USA, 2000, pp. 12e15. [7] J. Zhang, W. Prationo, L. Zhang, Z. Zhang, Computational fluid dynamics modeling on the air-firing and oxy-fuel combustion of dried victorian brown coal, Energy Fuels 27 (2013) 4258e4269. jek, Validation of an effervescent spray model with secondary [8] J. Broukal, J. Ha atomization and its application to modeling of A large-scale furnace, Appl. Therm. Eng. 31 (2011) 2153e2164. [9] R.I. Singh, A. Brink, M. Hupa, CFD modeling to study fluidized bed combustion and gasification, Appl. Therm. Eng. 52 (2) (2013) 585e614. [10] B. Rahmanian, M.R. Safaei, S.N. Kazi, G. Ahmadi, H.F. Oztop, K. Vafai, Investigation of pollutant reduction by simulation of turbulent non-premixed pulverized coal combustion, Appl. Therm. Eng. 73 (1) (2014) 1222e1235. [11] A.C. Benim, B. Epple, B. Krohmer, Modelling df pulverised coal combustion by a eulerian-eulerian two-phase flow formulation, Prog. Comput. Fluid Dyn. 5 (6) (2005) 345e361. [12] E. Scheepers, Y. Yang, A.T. Adema, R. Boom, M.A. Reuter, Process modeling and optimization of a submerged arc furnace for phosphorus production, Metallurgical Mater. Trans. B 41 (B) (2010) 990e1005. [13] B. Magnussen, B. Hjertager, On the structure of turbulence and A generalised eddy dissipation concept for chemical reaction in turbulent flow, in: AIAA Aerospace Meeting, St. Louis, USA, 1981, pp. 12e15. [14] D.L. Marchisio, A.A. Barresi, Understanding coal gasification and combustion modeling in general purpose CFD code, IEEE Trans. Plasma Sci. 58 (2003) 3579e3587. [15] M.M. Baum, P.J. Street, Predicting the combustion behavior of coal particles, Combust. Sci. Technol. 3 (5) (1971) 231e243. [16] L. Hookyung, S. Choi, B.J. Kim, Coal gasification and combustion modeling in general purpose CFD code, J. Korean Soc. Combust. 15 (3) (2010) 15e24. [17] J. Zhang, M. Zhang, J. Yu, Extended application of the moving flame front model for combustion of a carbon particle with a finite-rate homogenous reaction, Energy Fuels 24 (2010) 871e879. [18] E. Gomez, D. Rani, C. Cheeseman, D. Deegan, M. Wise, A. Boccaccini, Thermal plasma technology for the treatment of wastes: a critical review, J. Hazard. Mater. 161 (2e3) (2009) 614e626. [19] I. Istadia, Y. Bindarb, Improved cooler design of electric arc furnace refractory in mining Industry using thermal analysis modeling and simulation, Appl. Therm. Eng. 73 (1) (2014) 1129e1140.
0.2 0 0
1
2
3
4
5
6
Data Point Fig. 10. Proportional gas temperature differences for each data point. Comparison of the current simulation and Zahng et al. [7].
In this study, the thermal effects of the coal combustion and the electrode based arc radiation are modeled for a commercial EAF using the state of art simulation technique of CFD. Core temperature occurred with the combustion in the jet and the temperature distribution on the slag surface has been analyzed in detail. The variation of the CO and CO2 which are occurred by the combustion reactions and diffusion from the slag surface are given numerically and graphically. The effect of the radiation on the temperature distribution inside the EAF and on the slag surface is examined. The total and the average temperature distribution on the slag surface is around 1770 K. It is important to keep scrap metal at the fluid phase for the better quality. The average temperature at the slag surface is 1880 K while the minimum temperature is 1770 K. This temperature distribution meets the requirements. The developed model can help us to increase the thermal efficiency of an EAF by testing at the design level. It was determined that the model is fast, reliable, low cost and able to give detailed results. This model gives us an economic alternative against the costly experimental studies and also gives realistic and detailed results against the zero dimensional calculations which have the less consistency. It is seen that the major heat transfer mechanism is the radiation predictably. As a result of the study the heat transferred to the metal slag surface is very poor in comparison with the heat transferred to the EAF walls and the roof. The radiation heat loss from the EAF's wall and the roof is considerable.