Experimental and modelling study of low-nox industrial burners A. Frassoldati 1, A. Cuoci 1, T. Faravelli 1, E. Ranzi 1, D. Astesiano 2, M. Valenza 2, P. Sharma 3 1. Dipartimento di Chimica Materiali e Ingegneria Chimica, Politecnico di Milano, Italy 2. Danieli Centro Combustion, Corte dei Lambruschini, Genova, Italy 3. Chemical Engineering Department, IIT Madras, India Abstract This paper is concerned with the application of a rapidly developing combustion technology [1] that has been given different names: high-temperature air combustion (HiTAC [2]), MILD [3-5], or FLAMELESS combustion [6]. Aim is to present and discuss the effects of burner geometry and fuel/air inlets on NOx formation in an industrial burner using experimental and modeling activities. For this purpose a commercial burner (DCC MAB) was studied. We aim to optimize its geometry following the guidelines of flameless combustion where recirculation of exhaust flue gases dilutes the entering air, hence the concentration of oxygen within the reaction zone is much lower compared to ordinary flame combustion. The burner is simulated using the CFD code FLUENT 6.3.26 and then post-processed by means of a Kinetic Post-Processor (KPP) using a detailed kinetic scheme. The successful prediction of NOx supports the use of this tool to improve the design and assist the scale-up of the burner, with a particular attention to pollutants formation. 1. Introduction High temperatures are common in combustion chambers and to improve the efficiency a part of this heat is extracted by pre-heating the inlet air using the exhaust gases of the chamber. This results in high flame temperature which increases the formation of NOx. With the growing concern on Environmental policies a limit is put on amount of NOx emission with national and international regulation decreasing it gradually. To achieve this objective, efforts have been made in the last few years to develop innovative technologies such as flameless combustion. The underlying principle used are a) a strong (high-momentum) air jet that is surrounded by a number of weak (lowmomentum) fuel jets (typically two) b) a central fuel jet and a number of air jets positioned in the relative vicinity of the central fuel jet. The fundamental purpose behind this is the dilution of inlet air with inert combustion products before combustion starts. For this combustion regime temperature needs to be maintained above the self ignition temperature of the fuel, which is about 850 C natural gas. It is therefore generally necessary to have burners working in the flame regime which are capable of pre-heating the furnace to that temperature. Moreover, flameless regime provide a more uniform heat flow and temperature profile in the combustion chamber, which is certainly advantageous for heating and thermal treatment ovens. The burner, however, should allow a wide flexibility in terms of charge regulation and thermal profile. It is well known that in combustion systems (burners) the NOx formation is mostly thermal NOx which is influenced both by burner geometry, and operating factors like air excess, preheating temperature of combustion air and operating temperature of the oven. For having the possibility to use the burner in both flame and flameless modes it is required to use of control valves, installed on the feed lines of the fuel or hot air, for distributing the flow IX-2, 1
31st Meeting on Combustion rates according to the desired combustion regime (flame or flameless). This solution, however, can have high cost when the valves are installed on the hot air feeding lines. Present activity aims at a simple, economical and feasible solution of the above mentioned drawback complying with the low emission restrictions for a wide range of operating conditions. In view of the above objectives, a modification in existing traditional flame burner (DCC MAB) is suggested. A new gas lance, capable of injecting the fuel at high velocity, is installed in the burner. Through this modification this burner can now operate both in the flame mode (i.e. to heat-up the furnace) and in flameless mode, with low emissions of NOx, by acting on the distribution system of the (cold) fuel. When the fuel is injected according to the flameless technology, it is possible to obtain a sufficient degree of mixing between fuel, preheated air and exhaust gases so that combustion takes place in a more diluted manner. The low O 2 concentration reduces the reactivity thus limiting the formation of temperature peaks and correspondingly the formation of nitrogen oxides (thermal NOx) is inhibited. During the flame mode, the fuel is injected into the combustion chamber by means of the traditional central fuel inlet, while the burner can be operated in flameless mode using the high velocity gas lance, simply located inside one of the secondary air ducts. In this way, jets of combustible gas and preheated air meet at a pre-established and controlled distance A typical representation [6,7] of the various combustion regimes involves T vs. K V diagrams as shown in Fig. 1. Such a diagram identifies four different regions: traditional combustion (A), transition (B), flameless (C) and no combustion (D). Temperature [ C] 1600 1200 800 400 A B C D Table 1. Operating conditions Flow Rate Temperature Natural Gas 199 Nm 3 /h 20 C Air 1991 Nm 3 /h 450 C Furnace Temperature 1100 1300 C 0 0 2 4 6 8 K V Fig. 1. Combustion regimes stability limits (schematic)(adapted from [6,7]). Exhaust gas recirculation rate is defined [6] as the ratio m& Exhaust KV = m& + m& Fuel Air Cavigiolo et al. [7] and Derudi at al.[8], showed that for pure methane a K V value higher than 3.5 is necessary to reach the mild combustion regime. Operating conditions studied in this paper are summarized in Table 1. 2. The Danieli Centro Combustion (DCC) test facility The Danieli Centro Combustion test furnace (Fig. 2) is set to install radiant burners and side burners up to 3 MW of thermal power. Burner design is tested on 1:1 scale as design in function of the specific features required by the process. The combustion air can be pre heated up to 500 C by dedicated heat exchanger. IX-2, 2
Italian Section of the Combustion Institute The fuel gas system is designed for preparation of different mixtures of Natural gas or fuel gas fed to chamber. A series of thermocouple helps in regulation and monitoring of temperature profile inside the chamber. Sampling probes situated at different positions analyses the exhaust gas composition (O 2, CO, NOx ). Air & gas flow ratio, as well as the combustion air temperature and furnace pressure, are controlled by a PLC and the supervision/recording of the process is performed by a PC. Details on this furnace and a comparison between measurements and modeling for hydrogen-containing fuels can be found elsewhere [9] Fig. 2. Danieli Centro Combustion research facility furnace. 3. Burner Geometry - CFD modeling The DCC MAB burner geometry and furnace is shown in Fig. 3. The burner is equipped with a double fuel inlet (a central gas inlet and a gas lance for flame and flameless combustion, respectively), i.e. it is able to operate in both flame and flameless configurations. Symmetry Plane Fig. 3. View of the furnace (left) and mesh detail on the plane of symmetry(right). Turbulence was modeled by the RANS approach, using the standard k- closure model. A sensitivity analysis on the turbulence model showed a minor impact of this assumption on the predicted air distribution and velocities. For the spatial resolution the First-Order Upwind Scheme was adopted. The SIMPLE [10] algorithm is used for the coupling of pressure and velocity. The radiating heat transfer of the flame is calculated with the DO model [10]. The furnace was simulated in the steady-state regime using a 3D mesh with 1 million of cells which describes only half of the physical domain to take advantage of the symmetry of the system. IX-2, 3
31st Meeting on Combustion Fig.4 compares Temperature and velocity fields (in the symmetry plane) for the two operating conditions: as evident, the maximum temperature is reduced of 200 K when the burner is operated in flameless regime. T [K] flame V [m/s] flame flameless flameless Fig. 4. Temperature (left) and velocity (right) fields calculated according to the operating conditions of Table 1. Figure 4 also shows the flow field obtained in the flameless mode. Due to the high velocity of the jet, the Mach number exceeds 0.3, for this reason we have considered the compressible flow properties. For compressible flows, the ideal gas law is written in the following form: p+ p op ρ = RT where p op is the operating pressure and p is the local static pressure relative to the operating pressure. Dilation dissipation, which is generally neglected for incompressible flows, is modeled here using Sarkar s proposal (1990) [11]. In particular, in the k- model, the k equation is modified including the dilatation dissipation 2 term YM = 2 ρ ε Mt, where the turbulent Mach number, Mt, is 2 M t = k a and a is the speed of sound a = γ R T. On neglecting compressibility effects, the maximum flame temperature is underestimated by 50 K. This results is important for the successive calculation of NOx formation using the Kinetic Post-Processor. The description of the turbulence-chemistry interactions represents a critical aspect in modeling turbulent combustion. In this work the Finite-Rate Eddy Dissipation (FR-ED) model was used [12]. CH 4 reacts following a 2-step reaction mechanism, derived form the works of Westbrook and Dryer [13,14]. This kinetics schemes is shown in Tables 2 and 3. The role of the reverse reaction, which becomes significant at high temperatures, is to control the CO/CO 2 equilibrium concentration and the heat release intensity. Table 2: Reaction scheme. REACTION SCHEME 1 (a) CH 4 + 1.5O2 CO + 2 H2O 2step (b) 1 2 f,2r CO + O CO 2 2 2 IX-2, 4
Italian Section of the Combustion Institute Table 3: Constants used in the 1step and 2step approach. (Units are kmol-m-s-kj-k) Pre-Exponential Factor Activation energy Rate Exponent 2-step A 2 = 5.012x10 11 A 3f = 2.239x10 12 A 3r = 5.0x10 8 E a2 = 2.0x10 5 E a3f = 1.7x10 5 E a3r = 1.7x10 5 α( CH4 ) = 0.7, α( O2 ) = 0.8 α( O2 )=0.25,α( H2O )=0.5,α( CO )=1.0 α( CO2 ) = 1.0 4. KPP Results and Analysis The Kinetic Post-Processor (KPP) operates assuming the temperature and flow fields as predicted by the CFD and solves mass balances in the cells with detailed chemistry at fixed temperatures [15]. Two major features make this approach advantageous over the direct coupling of detailed kinetics and CFD. The first is the possibility to lump and group several cells and the latter is to fix the temperature inside the cells. CFD solution provides detailed flow, composition and thermal fields allowing the identification of the zones where the detail of the description can be reduced. The fixed temperature inside the cells reduces the high non linearity of the system, mainly related to the reaction rates and to the coupling between mass and energy balances. The calculations presented in this paper were obtained using about 50.000 reactors. Additional details on the KPP are reported elsewhere [15]. The KPP uses a detailed kinetic mechanism (C1C30704NOx) [16], containing about 85 species, to model combustion and pollutants formation in the gas phase. The results of the KPP analysis are briefly discussed in this section and compared to the experimental measurements. Fig. 5 shows the measured NOx emissions from the furnace, as a function of the furnace average temperature. As expected, the emissions increase as a function of the Furnace Temperature. [14] [14] [13] NOx [Arbitrary Units] 100 80 60 40 20 Air Temperature 450 C Flame combustion Flameless combustion 0 1100 1150 1200 1250 1300 Furnace Temperature [ C] Fig. 5. Measured NO emissions [a.u.] as a function of the temperature of the furnace for flame and flameless combustion ( 1% O 2 in the exhaust gases). Max Min T 293 2220 K NO 0 100 a.u. HCN 0 1 a.u. Fig. 6. Temperature, NO and HCN in the burner region of the furnace (flame mode). The emissions in flameless mode are almost half compared to the traditional flame. The NOx emissions, predicted at 1200 C, are in agreement with measurements. More importantly, the reduction ( 50%) of NOx emissions observed when the burner is operated in the flameless mode is well captured by the model. By analyzing the KPP results, it is possible to evaluate the role of the different NOx formation mechanisms. In the flameless configuration only 20% of the total NO is formed through the thermal mechanism, because of the relatively low temperature and residence time of the gases at high temperatures. Fig 6 shows the NO and HCN fields obtained from the flame mode. It can be observed, that maximum NO concentration is associated to the temperature hot spot, induced as a consequence of the modified secondary air inlet. HCN, which is an indicator of IX-2, 5
31st Meeting on Combustion the prompt-nox path is formed in the early stage of combustion, when CH 4 is consumed and forms CH i radicals. 5. Conclusions A turbulent methane burner is studied experimentally in flame and flameless combustion modes. It was modeled using a commercial CFD code assuming a 2-step kinetic mechanism. CFD results were post-processed using a numerical code called Kinetic Post Processor (KPP) which models turbulent reactive flows with detailed chemistry. Both experimental measurements and CFD simulations show that the modified burner in flameless mode reduces NOx emission by a factor of two. The computed results of NOx formation satisfactorily emulate the experimental emissions but are slightly underestimated. The proposed approach seems very encouraging as a solution to manage the complexity involved in the modeling of turbulent reactive flows using detailed kinetic models and can be applied for the design and scale-up of practical combustion systems. Of course, the reliability of model predictions is strongly dependent on the completeness and consistency of the original CFD simulation. Future research activity will involve the analysis of possible geometrical modifications and fuel distribution on the formation of NOx to optimize the burner also in the flameless operating mode. 6. References 1. R. Weber, A.L. Verlaan, S. Orsino and N. Lallemant, J. Inst. Energy 72 (1999), pp. 77 83. 2. H. Tsuji, A. Gupta, T. Hasegawa, M. Katsuki, K. Kishimoto and M. Morita, High Temperature Air Combustion, CRS Press, New York (2003). 3. M. De Joannon, A. Cavaliere, T. Faravelli, E. Ranzi, P. Sabia and A. Tregrossi, Proc. Combust. Inst. 30 (2005), pp. 2605 2612. 4. A. Cavaliere and M. De Joannon, Prog. Energy Combust. Sci. 30 (2004), pp. 329 366. 5. R. Weber, J.P. Smart and W. van der Kamp, Proc. Combust. Inst. 30 (2005), pp. 2623 2629. 6. J.A. Wünning and J.G. Wünning, Prog. Energy Combust. Sci. 23 12 (1997), pp. 81 94. 7. A. Cavigiolo, M.A. Galbiati, A. Effuggi, D. Gelosa, R. Rota, Combust.Sci.Technol. 175 (2003) 1347 1367. 8. M. Derudi, A. Villani, R. Rota, Proc. Combust. Inst 31 (2007) 3393 3400. 9. M. Valenza, A. Frassoldati, A. Cuoci, T. Faravelli, E. Ranzi, NOx formation in burners firing gases with low calorific value. An experimental and modeling study, paper 47. 6th International Symposium on High Temperature Air Combustion and Gasification, 17-19 October, HiTACG 2005 Essen (Germany), p 47_1-47_10. 10. FLUENT. Fluent Inc., FLUENT 6.3.26 User Manual (2007). 11. S. Sarkar and L. Balakrishnan. Application of a Reynolds-Stress Turbulence Model to the Compressible Shear Layer. ICASE Report 90-18, NASA CR 182002, 1990. 12. Frassoldati, A., et al., Chemical Engineering Science, 2005. 60: p. 2851. 13. Dryer F. K., Glassman I., Proc. Comb. Inst. 14 (1973) 987. 14. Westbrook C. K., Dryer F.K., Combustion Science and Technology 27 (1981) 31-43. 15. Cuoci, A., Frassoldati, A., Buzzi Ferraris, G., Faravelli, T., Ranzi, E., Int. J. of Hydrogen Energy, 32: 3486-3500 (2007) 16. http://www.chem.polimi.it/creckmodeling/kinetic.html IX-2, 6