Energy Procedia 14 (2012) 127 – 132
1876-6102 © 2011 Published by Elsevier Ltd. Selection and/or peer-review under responsibility of the organizing committee of 2nd International Conference on Advances in Energy Engineering (ICAEE).
doi:10.1016/j.egypro.2011.12.887
Energy
Procedia
Energy Procedia 00 (2011) 000–000
www.elsevier.com/locate/procedia
2011 2
ndInternational Conference on Advances in Energy Engineering (ICAEE 2011)
Simulation of High Temperature Air Combustion with
modified
Eddy-Break-Up
combustion model
Yaxin Su*, Cuiwu Chen, Along Su
School of Environmental Science and Engineering, Donghua University, 2999 Renmin Road North, Songjiang District, Shanghai, 201620, China
Abstract
The numerical modeling of High Temperature Air Combustion (HTAC) could correctly predict the flame and emission characteristics so far, however, the precision is not satisfactory. In this paper, a parameter-modified Eddy-Break-Up (EBU) model with a three-step reaction scheme was used to simulate the HTAC process in a furnace of 2m
×2m×6.25m. Reynolds stress model (RSM) is used for the computational fluid dynamics (CFD) simulation.
Comparison between experimental data from published papers, the original and modified EBU combustion models demonstrates that a smaller coefficient ‘A’ in the volumetric fuel consumption rate of the EBU model could improve the modelling precision. When ‘A’ is modified to be 1 other than its original value, 4, in the EBU model, the modeling could give improved results of the local distribution of CH4, temperature and NO emission.
© 2011 Published by Elsevier Ltd. Selection and/or peer-review under responsibility of [name organizer] Keywords: High Temperature Air Combustion; numerical simulation; modified Eddy-Break-Up (EBU) model;
1. Introduction
High Temperature Air Combustion (HTAC) technology was developed in 1990s and has been widely applied to many industries, e.g., iron and steel industry, mechanical manufacture, glass and ceramic kilt heating and power engineering. HTAC uses efficient regenerative burners to recover the heat of flue gases to preheat the combustion air above 1000°C, while the combustion is carried out in an atmosphere of low oxygen concentration as well as at high temperatures of the oxidizer, mostly above the auto-ignition temperature of the fuel. Therefore thermal nitrogen oxide (NO) is suppressed and high energy
* Corresponding author. Tel.: +86-21-67792552; fax: +86-21-67792522.
E-mail address: [email protected] ( Yaxin Su)
© 2011 Published by Elsevier Ltd. Selection and/or peer-review under responsibility of the organizing committee of 2nd International Conference on Advances in Energy Engineering (ICAEE).Open access under CC BY-NC-ND license.
Nomenclature
A empirical coefficient , A=4 ( EBU ) or 1 (modified EBU) B empirical coefficient, B=0.5
k turbulence kinetic energy, m2/s2 M molecular weight, kg/mol m mass fraction
n mole number
RF fuel consumption rate, kg/(m3.s) ρ Density, kg/m3
ε turbulence kinetic energy dissipation rate, m-2s-3 subscript
O ,F, P oxygen, fuel, combustion products
efficiency is achieved [1]. HTAC is also named as flameless oxidation (FLOX) [2], moderate and intense low oxygen dilution (MILD)[3] combustion. Both experimental and numerical research works on HTAC flame and emission have proved that HTAC has many advantages that are superior to conventional turbulent diffusion flames, e.g., a high flame stability with enlarged flame volume, a more uniform temperature filed and high radiative fluxes, very low NO emission, etc [4-10].
Modeling of HTAC in industrial furnace is important for design and operation management, however, it is usually difficult and complex, because the numerical work has to take into account both the combustion chemistry, NO formation chemistry, the turbulent flow and mixing between the fuel, preheated air and the flue gases in the furnace. Reynolds–averaged Navier–Stokes (RANS) approach based on standard k-ε turbulent model or RNG k-ε model or Reynolds stress model (RSM) is used for the computational fluid dynamics (CFD) simulation coupled with different combustion models. Mancini et al [6] reported their simulation by using standard k-ε turbulent model and three combustion models including Eddy-Break-Up (EBU) model with a two-step reaction scheme, the eddy-dissipation concept model with chemical equilibrium, and the pdf/mixture fraction model with nonadiabatic lookup tables. Their numerical results could predict reasonable NO emission, but failed to predict the structure of the weak fuel jet, the temperature and CH4, CO2 as well as CO concentrations. Orsino et al [7] used Eddy-Break-Up model, Eddy-Dissipation Concept model with chemical equilibrium, and a pdf/mixture fraction model with equilibrium to predict the high temperature air combustion process. But they failed in predicting the chemistry and temperature field in the fuel jet region. Yang and Blasiak [8] used Eddy-Break-Up model and standard k-ε turbulent model to calculate the HTAC of LPG and the predicted results of temperature and gas species are still have an obvious error when compared to experimental data. Yang and Blasiak [9] used RNG k-ε model and two combustion models, i.e., eddy-breakup model with a three-step chemical equilibrium, PDF model, to predict the HTAC process. They compared the calculated flame size and shape to a flame photograph and concluded that eddy-breakup model was suitable for the prediction. Validation of the temperature distribution, NO emission and gas components were not conducted.
In spite of the fact the current numerical simulation could correctly predict the flame and emission characteristics of HTAC, the precision is not satisfactory. In this paper, a parameter-modified EBU combustion was proposed to improve the simulation precision.
2. Numerical Model
The combustion in an industrial furnace is simulated by a combination of turbulent flow model and a suitable combustion model. Several authors have concluded that Eddy-Break-Up (EBU) model was suitable for the prediction the HTAC performance although there were errors. In the present work, EBU model is used with a modification to the empirical parameter.
2.1. Eddy-Break-Up Combustion model and parameter modification
The reaction rates of EBU combustion model were calculated from the Arrhenius rate expression. The volumetric fuel consumption rate is given by
⎟⎟
⎠
⎞
⎜⎜
⎝
⎛
−
=
P P O O F FS
m
B
S
m
M
A
k
R
ρ
ε
min
,
,
（1） Where F F O O OM
n
M
n
S
=
,
F F P P OM
n
M
n
S
=
,
A and B are empirical coefficients, A=4, B=0.5 in the original EBU model used by Yang and Blasiak [8,9]. The first two items in the brackets in equation (1) determine the local rate-controlling concentration, while the third is intended to inhibit the reaction where the temperature is low. The micro mixing time scale is taken to be the dissipation time scale of turbulent vortex, k/ε.The empirical coefficients, A and B are usually modified to improve the prediction. HTAC is supper-enthalpy combustion in nature. The preheated air temperature could be even higher than the fuel ignition point and the temperature in the furnace is much higher than the normal combustion process. Therefore, the coefficient B has little effect on the reaction rates because the constant B notes the inhibition of combustion by lower temperature. HTAC is carried out in very low oxygen condition and the combustion rate is usually more slowly than normal combustion processes. When the coefficient A decreases, the volumetric fuel consumption rate described by equation (1) will decrease. So the constant A could be set a smaller value for HTAC calculation. In the modified EBU model, A is modified to be 1 other than its default value, 4.
A three-step reaction scheme is considered in the present study. During the combustion process, the fuel is considered to convert first to CO and H2 in a lean oxygen condition and then CO/H2 mix with the oxygen and the flue gas in reaction zone due to the turbulent transportation to complete the conversion of CO/H2 to CO2/H2O.
CnHm+0.5nO2→nCO+0.5mH2 (R1)
CO+0.5 O2→CO2 (R2)
H2+0.5 O2→H2O (R3) 2.2. Turbulent flow, NO formation and radiation models
Reynolds stress model (RSM) is used to calculate the time-averaged Navier–Stokes equations. A full NO mechanism[10] including thermal and prompt NO, the NO formation via N2O intermediate mechanism and NO reburning models were used to predict NO emissions. Radiation was handled by Discrete Coordinates (DO) model. The radiation properties of flue gas are assumed to be of ‘grey body’ type and
can be temperature- or concentration- dependent, or both. A variable absorption coefficient model called weighted-sum-of-grey-gases model (WSGGM) was used to determine the variable absorption coefficient. 3. Results
The International Flame Research Foundation (IFRF) conducted experiment of HTAC of natural gas in an industrial furnace [6]. The injector of fuel and preheated air is showed in Fig 1(a) and the cross section of the furnace is showed in Fig 1(b). The furnace is 2m×2m×6.25m with one burner/injector in one of the side wall. The injector is made up of one preheated air jet in the center and four fuel jets surrounding the air jet. The diameters of the air jet and fuel jets are 124 mm and 10 mm respectively. The distance between the air jet and fuel jet is 280 mm as showed in Fig 1(a). The preheated air and the fuel are fed into the furnace at a velocity of 285 m/s and 100 m/s respectively. Table 1 lists the IFRF experiment conditions, which is also used as boundary conditions for the numerical calculation.
Fig 1 (a) Injector ; (b) Furnace cross section Table 1 Fuel, preheated air and flue gases in IFRR test
The computational fluid dynamics (CFD) code, FLUENT 6.3, is used to carry out the CFD calculation. The HTAC process of the IFRF experiment is simulated with the original and modified EBU models. Hexahedral cell is used to mesh the whole furnace and grid refinement is adopted near the fuel nozzle and combustion air nozzle. Fig 2 presents the mesh for CFD calculation. The grid contains altogether 520,000 cells for the furnace showed in Fig 1(b). The experimental data reported by IFRF is used to validate the present model.
3.1. CH4distribution in furnace Flow rate
(Kg/h) Temperature (K) Enthalpy (MW) Components （% vol）
Fuel 47 298 0.58 CH4: 87.8%, C2H6: 4.6%, C3H8: 1.6%, C4H10: 0.5%, N2::5.5%
Preheated air 830 1573 0.35 O2: 19.5%, N2: 59.1%, H2O: 15%, CO2: 6.4%, NO:110ppm
Fig 3 presents the distribution of CH4, one of the major fuel components in furnace. It’s seen that the modified EBU model could give a better prediction compared to experimental data by Mancini et al [6]. The coordinate Y notes the direction in the furnace width and Z notes the direction of furnace length.
Radial CH4 at Z=0.73 m 0 2 4 6 8 10 0 0.2 0.4 0.6 0.8 Y (m) CH 4 (% ) experimental data EBU modifided EBU
Fig 2 Mesh used for simulation Fig 3 Calculated CH4 distribution in furnace
3.2. Temperature distribution
Fig 4 presents the simulated temperature distribution in the furnace. The results of modified EBU model agree better with the experimental data [6] than that of EBU model. When Z=0.15m, i.e. the zone very near the air/fuel jets inlet, modified EBU could give a temperature very close to measured data, while EBU model over-predicts the temperature when Y is between 0.1m to 0.28m,i.e, the zone between fuel jet and air jet where combustion happens when the fuel and preheated air meet. In EBU model, the parameter ’A’ takes 4, which leads to a fast reaction rate, while in the modified EBU model, the parameter ‘A’ takes 1, which decreases the reaction rate, leading to the improved prediction of both temperature and CH4 consumption in furnace. When Z is 0.73m, the modified EBU give a better prediction of temperature than EBU model.
temperature at Z=0.15 m 500 900 1300 1700 2100 0 0.2 0.4 0.6 0.8 Y (m) Te m pe ra tu re ( K ) Experimental data Modified EBU EBU temperature at Z=0.73 m 1200 1400 1600 1800 2000 0 0.2 0.4 0.6 0.8 Y (m) Te m per at ur e ( K ) Experimental data EBU modifided EBU
Fig 4 Temperature distribution in furnace (Y: width ; Z: length) (a)Z=0.15m; (b)=0.73m 3.3. NO emission
Fig 5 presents the simulated NO distribution in the furnace. In the front zone in furnace, e.g., Z=0.73 m, the simulated NO by modified EBU model agree with the experimental data [6] better than that by EBU model, although there is still an error. However, in the rear zone in furnace, e.g., Z=2.05 m, the modified EBU model could give a very precious simulation of NO compared to the tested data, while the error between the simulated and experimental data of NO by EBU model is rather obvious.
NO at Z=0.73 m 0 30 60 90 120 150 180 0 0.2 0.4 0.6 0.8 1 Y (m) NO ( pp m ) Experimental data Modified EBU EBU NO at Z=2.05 m 60 90 120 150 180 0 0.2 0.4 0.6 0.8 1 Y (m) NO (p pm ) Experimental data Modified EBU EBU
Fig 5 NO distribution in furnace (Y: width ; Z: length) (a) Z=0.73m; (b)=2.05m 4. Conclusion
The high temperature air combustion of natural gas in an industrial furnace was numerically modelled by a parameter-modified Eddy-Break-Up (EBU) combustion model with a three-step reaction scheme. Comparison between experimental data from published papers, the original and modified EBU combustion models demonstrates that a smaller coefficient ‘A’ in the volumetric fuel consumption rate of the EBU model could improve the modelling precision. When ‘A’ was modified to be 1 other than its original value, 4, in the EBU model, the modeling could give a better simulation results of the local distribution of CH4, temperature and NO emission.
Acknowledgements
The work was supported by Shanghai Natural Science Foundation (No.11ZR1401000), which is gratefully acknowledged.
References
[1] Tsuji H, Gupta A, Hasegawa T. High Temperature Air Combustion. New York: CRS Press; 2003.
[2] Wünning JA, Wünning JG. Flameless oxidation to reduce thermal NO-formation. Prog Energy Combust Sci 1997; 23: 81–94.
[3] Cavaliere, M. de Joannon, Mild combustion. Prog Energy Combust Sci 2004; 30: 329–66.
[4] Choi G, Katsuki M. Advanced low NOX combustion using highly preheated air. Energy Convers Mgmt 2001; 42: 639-52.
[5] Dormire J C. Benefits of applying regenerative burner technology to continuous reheat furnace. AISE Steel Technol. 2000; 77: 55-7.
[6] Mancini M, Weber R, Bollettini U. Predicting NOX emissions of a burner operated in flameless oxidation mode. Proc Combust Institute 2002; 29: 1155-63.
[7] Orsino S, Weber R, Bollettini U. Numerical simulation of combustion of natural gas with high-temperature air. Combust Sci Technol 2002; 168: 1–34.
[8] Yang W, Blasiak W. Numerical simulation of properties of a LPG flame with high-temperature air. Intl J Thermal Sci 2005; 44: 973–85.
[9] Yang W, Blasiak W. Numerical study of fuel temperature influence on single gas jet combustion in highly preheated and oxygen deficient air. Energy 2005; 30: 385–98.
[10] Yang W, Blasiak W. Mathematical modeling of NO emissions from high-temperature air combustion with nitrous oxide mechanism. Fuel Process Technol 2005; 86: 943–57.
