politechnika gdańska - Pomeranian Digital Library
Transkrypt
politechnika gdańska - Pomeranian Digital Library
POLITECHNIKA GDAŃSKA WYDZIAŁ MECHANICZNY KATEDRA TECHNIKI CIEPLNEJ Rozprawa doktorska Weihong Yang EXPERIMENTAL AND MATHEMATICAL MODELING STUDY OF GAS COMBUSTION UNDER THE HIGHTEMPERATURE AND OXYGEN-DEFICIENT OXIDIZERS Promotor pracy: Prof. dr hab. inż. Jan Stąsiek Gdańsk 2008 Pragnę złożyć serdeczne podziękowanie promotorowi Profesorowi Janowi Stąsiekowi za życzliwą pomoc, konsultancje oraz ogromną cierpliwość podczas realizacji pracy. Dziękuję również Profesorowi Włodziemierzowi Błasiakowi za umożliwienie pobytu naukowego w Royal Institute of Technology (KTH), Sztokholm, Szwecja i uczestniczenia w pracach badawczych realizowanych w laboratoriach Katedry Wymiany Ciepła i Pieców Przemysłowych (Division of Heat and Furnace Technology). TABLE OF CONTENTS List of Figures iv List of tables viii Nomenclature ix 1. 1 2. 3. Introduction 1.1 Motivation 1 1.2 Thesis 4 1.3 Objective 4 1.4 Scope of the work 4 State-of-the-art 6 2.1 History of the HiTAC 6 2.2 Study of single jet flame 9 2.3 Study of a semi-industrial furnace with HiTAC burners 14 2.4 Flame volume, length and entrainment 17 2.5 NO emissions 17 Methodology 20 3.1 Experimental facility 20 3.1.1 Single jet experimental facility 20 3.1.2 Classification of the HiTAC 22 3.1.3 The HiTAC test furnace 26 3.1.4 Measurement program 30 3.2 Mathematical model for HiTAC 31 3.2.1 turbulent combustion model 31 3.2.2 NOx modes 33 3.2.2.1 Thermal NO 34 3.2.2.2 Prompt-NO 35 3.2.2.3 N2O –intermediate Nox 36 3.2.2.4 NO reburning 38 3.2.2.5 NO source term and turbulence-chemistry interaction 39 3.2.3 Other models 40 i 3.3 Developed concept for description of HiTAC characteristics 4. 5. 6. 7. 40 Cross-flow single jet flame study 45 4.1 Study of mathematical modelling 45 4.2 Study of Mathematical models 49 4.3 Mean Residence Time and Flame Peak Temperature 52 4.4 Gas Temperature Uniformity Ratio 53 4.5 Conclusions 55 Flame volume and length study in co-flow single jet flame 56 5.1 Flame appearance 57 5.2 Effect of oxygen concentration 57 5.3 Effect of oxidizer temperature 63 5.4 Effect of fuel temperature 64 5.5 Effect of fuel flow rates 66 5.6 Effect of fuel nozzle diameter 67 5.7 Scaling analysis 68 5.8 conclusion 72 Flame entrainment study in co-flow single jet flame 73 6.1 Study Cases 74 6.2 Effects of the oxygen concentration of the oxidizer on entrainment 76 6.3 Effects of the temperature of the oxidizer on entrainment 83 6.4 Effects of the fuel flux on entrainment 84 6.5. Effects of the buoyancy 85 6.6. Global field behaviour of the entrainment 90 6.7 Conclusions 94 Semi-industrial furnace with HiTAC burner study 95 7.1 Experimental measured and verification of mathematical modelling for HiTAC furnace with one-flame burner 95 7.1.1 Energy Balance 95 7.1.2 Temperature Field 97 7.1.3 Gas Species 98 7.1.4 Features of Combustion and Flow of HiTAC 101 7.1.5 Heat Transfer Elevation 108 7.1.6 Simulation of a Moving Slab 110 7.2 Study of the HiTAC furnace with a twin-flame HiTAC burner 111 ii 7.2.1 Experimental and verification of the modelling prediction Heat transfer elevation 7.2.2 Effect of flame configurations 114 7.2.3 Effect of excess air ratio 117 7.2.4 Effect of Fuel/Air injection momentum ratio NO emission 119 7.3. OPTIMAL DESIGN OF A HiTAC FURNACE 120 7.3.1 Flame Entrainment Ratio 120 7.3.2 Optimal Design of HiTAC Furnace 124 7. 4 CONCLUSIONS 8. 111 NOx formation and destruction mechanism study 129 131 8.1 Experimental and verification of NO emissions models 131 8.2 Effects of NO model on the NO formations 134 8.3 NO emissions From Coflow Gas Jet Combustion Study 139 8.4 Conclusion 143 9. Thermodynamics investigation of the HiTAC combustion 144 10. Mainly Conclusion of the thesis 149 11. References 151 iii List of Figures Figure 3.1 Schematic of the combustion chamber for single fuel jet test furnace (a)Cross flow (b) Co flow 21 Figure 3.2 Scheme of one-flame HRS 24 Figure 3.3 Scheme of two-flame HRS 25 Figure 3.4 Two-flame regenerator system firing configuration for uniformity temperature profile in the furnace (F means Fuel, A means air and F.G. means flue gas) (a) single-side firing configuration (b) stagger firing configuration(c) counter firing configuration 25 Figure 3.5 HiTAC test furnace and burner (a) HiTAC test furnace at KTH, (b) Configuration of HiTAC test furnace with one-flame HRS Figure 3.6 Configuration of the HiTAC test furnace with inlet arrangements 27 28 Figure 3.7 Configuration of HiTAC test furnace with two-flame HRS (a) Two-flame HRS (b) Top view of HiTAC test furnace and test positions 28 Figure 3.8 Top view of HiTAC test furnace and test positions 29 Figure 4.1 3D-computational domain and meshes of HTAC test furnace 46 Figure 4.2 Predicted temperature profiles for 10% oxygen in the air preheated upto 1041K and at fuel inlet temperature equal to 473 K. (a) PDF model; (b)EBU model 47 Figure 4.3 Predicted distributions of oxidation mixture ratio for 10% oxygen in the air preheated upto 1041K and at fuel inlet temperature equal to 473 K. (The same scale as Figure 4.2) 47 Figure 4.4 Predicted gas temperature profiles for 10% oxygen in the air preheated upto 1041K and at fuel inlet temperature equal to 288 K. A=4.0, (b) A=2.0, (c) A=1.0 48 Figure 4.5 Predicted distributions of the oxidation mixture ratio for 10% oxygen in the air preheated upto 1041 K and at fuel inlet temperature equal to 288 K. (a) A = 4.0, (b) A = 2.0, (c) A = 1.0 Figure 4.6 Rf versus oxygen concentration for various air and fuel temperatures (K) 49 50 Figure 4.7 Rf versus preheated air temperature for oxygen concentration and fuel temperatures (K) 51 Figure 4.8 Mean residence time (s) versus oxygen concentration and combustion air and fuel temperature (K) 52 iv Figure 4.9 Peak flame temperature (Tmax) versus oxygen concentration for various combustion air and fuel temperature(K) 53 Figure 4.10 Temperature uniformity ratio, Rtu versus oxygen concentration for various air and fuel temperatures(K) 54 Figure 5.1 Flame appearance for the LPG jet co-flowing vertically with hot and oxygen depleted flue gas. Fuel nozzle diameter 0.5 mm. Fuel jet velocity equals to 25.5 m/s. Flue gas velocity equals to 0.98 m/s 57 Figure 5.2 Flame apparence versus oxygen concentration To = 1173 K, dF = 5.E-4 m, QF = 0.01 g/s, TF =299K 58 Figure 5.3 Predicted flame shape and size for different oxygen concentration (To = 1173 K, dF = 5.E-4 m, QF = 0.01 g/s, TF =299K) 59 Figure 5.4 Length and volume of HiTAC flame versus oxygen concentration.(To =1173 K. dF = 5.E-4 m. QF = 0.01 g/s. TF =299K) 60 Figure 5.5 Predicted temperature profiles for different oxygen concentration.(To =1173 K. dF = 5.E-4 m. QF = 0.01 g/s. TF =299K) 61 Figure 5.6 Flame maximum temperatures versus oxygen concentration.( To =1173K. dF = 5.E-4 m. QF = 0.01 g/s. TF =299K) 62 Figure 5.7 Flame maximum temperatures versus air temperature.( [O2]=12.8%. dF = 5.E4 m. QF = 0.01 g/s. TF =299K) 62 Figure 5.8 Length and volume of the HiTAC flame versus oxidizer temperature. ([O2]=12.8%. dF = 5.E-4 m. QF = 0.01 g/s. TF =299K) 64 Figure 5.9 Length and volume of the HiTAC flame versus fuel temperature. [O2]=10%. dF = 5.E-4 m. QF = 0.01 g/s. TO =1173K 65 Figure 5.10 Length and volume of the HiTAC flame versus fuel firing rate.(TO =1173K. dF = 5.E-4 m.) 66 Figure 5.11 Length and volume of the HiTAC flame at different cases ( [O2]=10%. dF = 5.E-4 m. TO =1173K) 67 Figure 5.12 HiTAC Flame length and volume versus diameter of fuel nozzle. ([O2]=10%. TO =1173K. TF=299K QF =0.01 g/s) 68 Figure 5.13 HiTAC Flame length and volume correlated with the flame Froude number 70 Figure 6.1 The variation of the entrainment rate along the axial direction from fuel nozzle tip at different oxygen concentrations in the oxidizer 76 v Figure 6.2 Entrainment rate change versus the oxygen concentration and the characteristic temperature ratio (T f / To ) 0.5 81 Figure 6.3 The variation of the entrainment rate along the axial direction from fuel nozzle tip at different preheated temperature of the oxidizer 83 Figure 6.4 The variation of the entrainment rate along the axial direction from fuel nozzle tip at different preheated temperature of the oxidizer 85 Figure 6.5 The local influence of buoyancy as a function of the axial distance normalize by a) (x/d*), and b) the flame height in log scale for difference cases 89 Figure 6.6 Dimensionless entrainment rates v.s. ratio of axial distance from the nozzle tip to flame height 91 Figure 7.1 Temperature distribution on the side wall of the furnace at x = 0.8 m, and y = 0.3m 98 Figure 7.2 Predicted and measured O2 profiles in the furnace x = 0, z = 0.3 m (b) x = 0, z = 0.6m (c) x = 0, z = 1.2 m 99 Figure 7.3 Predicted and measured CO profiles in the furnace (a) x = 0, z = 0.3 m , (b) x = 0, z = 0.6 m, (c) x = 0, z = 1.2 m, (d) x = 0, z = 2.15 m 100 Figure 7.4 Predicted velocity vectors at a cross section through the fuel and one of the air inlets in HiTAC mode [m/s] 102 Figure 7.5 Predicted temperature profile at a cross section through the fuel and one of the air inlets at HiTAC mode [K] 103 Figure 7.6 Predicted and measured flame shapes and volumes shown by the oxidation mixture ratio RO=0.99 (a) Predicted (b) Estimated by experimental data 105 Figure 7.7 Predicted flame shape Conventional burner 107 Figure 7.8 Predications of heat flux absorbed by the charge along central line on the surface of sink Case 0: Test furnace with a one-burner HRS without any charge or heat sink. Case 1: Test furnace with conventional turbulent jet flame without any charge or heat sink. Case 2: Test furnace with a one-burner HRS and with a stationary heat sink whose surface temperature is equal to 20OC and constant, Case 3: Test furnace with conventional turbulent jet flame with a stationary heat sink whose surface temperature is equal to 20OC and constant.Case 4: Test furnace with a one-burner HRS with a moving steel slab which initial surface temperature is equal to 20OC. vi 109 Figure 7.9 Predicted temperature distribution of moving slab with one-flame HRS 111 Figure 7.10 Predicted and measured CO profiles in the upper part of the furnace (a) x=0, z=1.7m, (b) x=-0.475 m, z=1.4m, (c) x=0., z=0.75m 113 Figure 7.11 Predicted and measured O2 profiles in the upper part of the furnace at x=-0.475 m, z=1.4m 114 Figure 7.12 Temperature profiles (K) at various firing locations (a) Counter mode (b) Single-side mode, (c) Stagger mode 114 Figure 7.13 Flame shapes shown by the oxidation mixture ratio RO=0.99 for difference firing modes (a) Counter mode (b) Single-side mode (c) Stagger mode 116 Figure 7.14 Effect of firing configurations on NO emissions 117 Figure 7.15 NOx emission vs excess air ration during the combustion of counter mode 118 Figure 7.16 NO variations with the ratio of the fuel/air injection momentum 120 Figure 7.17 Flame entrainment ratio vs. excess air ratio (ALFA) 121 Figure 7.18 Excess air ratio vs. flame occupation coefficient and peak temperature 122 Figure 7.19 Effect of fuel temperature and flame locations on flame entrainment ratio 123 Figure 7.20 The scheme of structure of the flame zone and recirculation in an enclosed combustion chamber 125 Figure 7.21 Distributions of Temperature and Flame entrainment ratio along the direction of the furnace width 128 Figure 8.1 Effects of excess air ratio on NO emission 132 Figure 8.2 Effect of excess air ratio on maximum temperature 133 Figure 8.3 Predicted NO contours (mass%) at the cross-section through fuel and one air nozzle in the test furnace at excess air ratio (λ) equal to 1.04 and 1.09 with and without N2O-route 136 Figure 8.4 Predicted NO contours (mass%) at the cross-section through fuel and one air nozzle in the test furnace at excess air ratio (λ) equal to 1.15 and 1.25with and without N2O-route 137 Figure 8.5 Temperature distribution at the cross-section through fuel and one air nozzle in the test furnace at different excess air ratio (λ) 138 vii Figure 8.6 Effect of oxygen concentration on NO emissions at an air temperature of 1173K and fuel temperature of 299K for gas jet combustion in the co-flow test 141 Figure 8.7 Effect of preheated air temperature on NO emissions at a fuel temperature of 299K for the case of gas jet combustion in the co-flow test 142 Figure 8.8 Effect of fuel flowrate on NO emission at an air temperature of 1173K and a fuel temperature of 299K for gas jet combustion in the co-flow test 142 Figure 9.1 Enthalpy-temperature diagram for CH4 mixture with different oxygen level at the stoichiometric 145 Figure 9.2 Excess enthalpy in log versus inlet temperature at different oxygen concentration for a methane/oxygen/nitrogen mixture 146 viii List of Tables Table 3.1: Arrhenius kinetic coefficients used for nitrous oxide mechanism reactions Units:cal,mol,cm3,sec 37 Table 6.1 Values of Variables of Numerical Studies 74 Table 6.2 Summary of the results for present study and previous works 78 Table 7.1 Fuel characteristic and burner operating conditions 96 Table 7.2 Energy balance from measurements and from modelling data (Reference temperature: 298K) 97 ix Nomenclature A - empirical coefficient Acr - criteria area of flue gas recirculation Ar - area of flue gas recirculation B - empirical coefficient C - pre-exponential factor of chemical reaction Ce - entrainment coefficient CHiTAC - Richardson correction ratio for the HiTAC combustion Cp - specific heat, kJ/(kg.K) D - diameter, m d* - equivalent diameter defined ds - source diameter, m Ea - activation energy, kJ/mol Fb - the buoyancy force, N Fm - the initial inertia force of the fuel nozzle, N. Fr - Froude number fs - the stoichiometry g - gravitational acceleration, m/s2. k - turbulent kinetic energy, m2/s2 ka - absorption coefficient of flue gas L - flame length, m L* - dimensionless flame length M - molecular weight (kgmol-1) m - mass fraction M molecular weight (kgmol-1) m* - non-dimension entrainment mass flux m0 - initial jet mass flux, me - the jet mass flux, n - stoichiometric coefficient (number of moles) qc - local heat release of difference fuel species, (kW/m3) Qf - the heat release in flame zone, (kW). qFHR - average Flame Heat releasing ( kW/m3) R - the universal gas constant, kj/kmol.K x r - the density weighted velocity ratio between the jet and coflow, (( ρ F u F2 ) /( ρou o2 )) 0.5 R - universal gas constant, (Jkmol-1K-1) r - radial distance to the flame centreline within the flame, m Re - Reynolds number REBU - fuel consumption rate for eddy break up model (kgm-3s-1) Rent - entrainment rate Rfe - flame entrainment ratio Rflame - radial distance from the flame centreline to the flame board, m. RFOC - furnace occupation coefficient RHiTAC - rate of high-temperature and oxygen deficient (HiTCA) combustion, kg/m3/s Ri - Richardson number RKIN - Arrhenius reaction rate (kgm-3s-1) RKIN - Arrhenius reaction rates, kg/m3/s Ro - oxidation mixture ratio Ro - oxidation mixture ratio Rtu - furnace temperature uniformity ratio S - stochiometric air to fuel mass ratio, T - temperature (K) ui - velocity in the i cell number, m/s v - velocity Vf - flame volume (m3). VF - furnace volume ( m3) vj - the species rate exponents, x - the downstream distance from the virtual origin, m. Z - horizontal distance from burner face, (m) β - temperature exponent δ - jet width, m ξ - the nondimensional streamwise coordinate ξ’ - correction of the nondimensional streamwise coordinate ξHiTAC - correction nondimensional streamwise coordinate for HiTAC combustion ρ - density, kg/m3 υ - the specie rate exponent u - average velocity, m/s T - average temperature in furnace (K) xi ε - turbulent kinetic energy dissipation,m2/s3 [O2] - oxygen concentration, % ∏ ( x ) - denotes the product of all xj. j β - temperature exponent of chemical reaction ΔT f - the characteristic temperature rise resulting from combustion, K Subscripts 0 - initial state ∞ - ambient a - air ad - adiabatic temperature c - species f - flame F - fuel F - Furnace Fl - fuel i - calculation cell number Max - maximum values O - oxygen O, o - oxidizer out - final state P - product r - recirculation flue gas s - source stoic - stoichiometry xii Abstract The combustion proceeds 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, spread out many new features, such as significantly higher flame stability at all fuel-air (including very lean mixtures) a larger flame volume, a uniform temperature distributions, higher heat transfer, low NOx emission. These features have been demonstrated in a lot of practice applications, either with air as oxidizers combing a modern regenerative system, or with pure oxygen as oxidizers. A comprehensive view of these works has been provided, in regard to the fundamental differences in the thermal, chemical and fluid dynamics characteristics of the flame. However, since the high temperature and low oxygen deficient are the unique parameters differ from conversations combustion, there is little knowledge of the quantitatively effects of them on the flame properties. The generally objectives of this work are theoretical analysis and development of new combustion concepts. In particular, • Quantitative study of the flame properties, attention is focused on the volume, length, and entrainment of an ‘invisible’ flame. • NOx formation and destruction mechanisms during new combustion conditions. The investigation has been performed using a single fuel jet flame facility with cross and co-flowing, and a semi-industrial furnace equipped with High-temperature Air Combustion-HiTAC burners. Experimental, numerical and theoretical analyzing investigations are adopted. In this work, a ‘chemical’ flame volume and ‘chemical’ flame length were used to describe this ‘invisible flame’. Results from single jet flame study show that: • Flame length increases with either the decrease of oxygen content, or increase of oxidizer temperature, or decrease of fuel temperature. Furthermore, the flame length is independent of the fuel flow rate and the diameter of the fuel nozzle for the studied cases. xiii • Flame volume increases either with the decrease of oxygen content and increase of oxidizer temperature, or with the reduction of fuel temperature, or with the increasing in fuel firing rate. Flame volume depends very much on the oxygen concentration in the oxidizer. • Influences of high temperature and low oxygen concentration in the oxidizer on the flame Froude number, Frf were examined. Regimes of momentum- or buoyancycontrol, were determined on the assumption that oxidizer temperature and oxygen concentration are changeable. A simple correlation of the ‘flame’ length and volume with flow parameters has been derived in terms of a flame Froude number for momentum-buoyancy transition jet flame under the HiTAC condition. The criteria constants of the dimensionless flame volume V* and the dimensionless flame length L* to assess momentum– or buoyancy–control flame are given. Additionally, the entrainments of this ‘invisible’ flame have been numerically and theoretical studied. Conclusions are: • The uniformity of the heat release in reacting jets has strong effect on the flame entrainment. More uniform the heat release, larger the entrainment. The effect of heat release reduces the entrainment in the near field of the reacting jets with the same factor of the characteristic ratio (Tf/To)0.5. • The entrainment increases as the oxygen concentration is decreased. Furthermore, the entrainment is independent of the fuel flow rate and the preheated temperature of the oxidizer for the investigated temperature range (1073-1573K). • The effect of the oxygen concentration and preheated temperature of the oxidizer on buoyancy was examined. A correction Richardson coordinate, where the effect of the oxygen concentration (stoichiometric ratio) is included, was derived to describe the local influence of buoyancy force along the chemical flame length under the high temperature and oxygen deficient oxidizer condition. It can be concluded that the buoyancy force increases with the reduction of the oxygen concentration in the oxidizer. • The global behaviour of the entrainment was revealed. The entrainment of jet flames can be identified as two regimes: (a) the near field where entrainment coefficient is positive; and (b) the far field where entrainment coefficient is negative. Corrections of entrainment rates were derived in terms of a Frf number for momentum-buoyancy xiv transition jet flame under the high temperature and low oxygen concentration oxidizer condition. Furthermore, the maximum entrainments along the flame length are estimated Further on, the benefits of HiTAC technology are quantitatively demonstrated by mathematical models. These benefits are: lower peak temperature, larger flame volume, more uniform thermal field, lower local firing rate, higher heat transfer, higher energy utilizing efficiency and lower combustion noise. NOx formation and destruction during this new combustion phenomenon has been studied numerically. It was found that the NO formation via N2O mechanism may be important. The approximate percentage of NO production by the nitrous oxide according to the Zeldovich and prompt mechanism varies from 5:95 at 10% oxygen concentration to 95:5 at 5% oxygen concentration. Finally, a thermodynamic analysis of combustion process shows that oxyfuel combustion is able to increases the available energy of the flameless combustion, thus higher combustion intensity. Additionally, the flameless oxyfuel burner does not need preheating the oxidizer, this extend the concept of the HiTAC. xv 1 Introduction 1.1 Motivation According to the Kyoto Protocol on Global Climate changes as the third conference of parties (COP3) to the United Nationals Frame conversion that developed countries should reduce their total emissions of greenhouse gases by at least 5% from the level of 1990 between the years 2008 and 2012. The European Union and its implementation recently decided to set a target of 20% CO2-reduction, and 20% renewable energy using. Fossil fuel, such as gas and oil is used in the various fields such as steel production, petroleum, electronics and machinery and the energy saving on these industrial processes will result in big effect for reducing global warming gases. In 2007 the fossil fuels constituted 80% of the global energy supply (35% petroleum + 24% coal + 21% natural gas) [1]. Further on, emission of NOx (NO,NO2, N2O, N2O4, N2O5,etc), which is known to be responsible for the destruction of ozone layer in the upper atmosphere, is another environment problem when fossil fuel is used. Therefore the energy saving and environment protection technologies should be developed and then applied as soon as possible in the practical fields. The most efficient saving energy technique for the combustion of fossil or derived fuels is either preheating combustion air /fuel by recovering flue gas heat or reducing the flue gas volume (oxygen enhancement in air ), or combining both of them. However, directly used of these methods will lead to high combustion temperature, thus a higher NOx emission High temperature air combustion-HiTAC technology can simultaneously reduce CO2 and nitric oxide emissions and reduce energy consumption. It has been now well recognized as the most advance combustion technique to achieve all requirements. The essence of the HiTAC is fuel combustion under the condition of high-temperature and oxygen and deficient atmosphere. A lower oxygen concentration during the combustion process can depress the flame temperature even at a higher preheated oxidizer /fuel, and a higher temperature of oxidizer can sustain the flame stability. The necessary temperature of the oxidizer can be decided according to the oxygen concentration in the oxidizer. 1 It has been now well recognized that HiTAC has advantages as: • Energy savings • CO2 reduction • NOx reduction • Uniformity of temperature profile • Enhancement of heat transfer • Lower combustion noise In the practical applications, the high-temperature of oxidizer can be obtained using a modern regenerative system, or mixing of cold oxidizer (air/oxygen) and the hot flue gas by a higher injection velocity of the oxidizer. For the fore method, the combustion oxidizer can be preheated by regenerative system directly, and for the latter, the oxidizer can be heated by stronger flue gas internal recirculation. Both technological solutions can be obtained high-preheated oxidizer temperature before combustion occurs. The temperature of the preheated oxidizer at the point before combustion occurs should be higher than the autoignition temperature of used fuel. The low oxygen concentration in the oxidizer can be obtained either by an external exhaust gas recirculation, or by an internal flue gas recirculation obtained by a high velocity injection of oxidizer and/or fuel. Gases and liquid fuel combustion with modern regenerative type of heat exchangers has been demonstrated in industrial furnaces, for example, reheating furnaces, heat treatment furnaces, melting furnace etc, for energy savings (about 30%), reduce pollutant’s emission, including CO2(about 30%), and reduce equipment size (about 25%) [2] Additionally, for oxygen enhancement or pure oxygen combustion, this kind of phenomenon is achieved by supersonic injection of oxygen, which forms a strong internal flue gas recirculation. This technology is called as ‘’flameless oxyfuel’’. Dramatically fuel energy savings have been successfully demonstrated in different types of furnaces along with uniform thermal field and pollution. High- temperature air combustion technology has already been applied in steel industrial, reported by Yasuda [3], Suzukawa [4] and Mori [5]. In all these projects very high fuel savings (sometimes as high as 60%), reduction of NOx (around 50%) and production 2 increase (by 20 -50 %) was achieved. Base on these factors, apart from steel industry, HiTAC were applied to melt aluminum or to incinerate odour, vapor gases for example in pulp and paper industry. Industrial applications of this so called “flameless” oxyfuel combustion are documented [610] and prove enormous future potential of this technology in various thermal processes for example soaking pits, box-, walking beam-, catenary furnaces [10]. This new combustion proceeds 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. The high-temperature (above the fuel’s auto ignition temperature) and oxygen deficient atmosphere are unique characteristics that differ from any other combustion. A more accurate definition would be that combustion is spread out in a large volume, which some refer to as ‘volumetric combustion’, resulting in uniform and lower temperature of such flame. The essence of these larger flame volume and lower flame temperature is the low temperature increase during the combustion process. This makes a combustion chamber more like a well stirred reactor. Again, the effect of heat release on the combustion characteristics is less in the case of a low temperature increase. For example, larger flame volume and gas entrainments are found during high temperature and oxygen deficient conditions. This works firstly concerns the studying of some thermal and physical properties of gaseous fuel flames, especial focus on the quantitative of ‘invislabe’ flame volume, length, and flame entrainment for both cross and co-flow of single jet flame. Further on, the flame properties in a semi-industrial HiTAC furnace with modern ceramic honeycomb regenerative burner have been studied. These concepts were used to describe the characteristics of the HiTAC in a practical application, which provides help for optimal design of a HiTAC furnace and burners. As well, the benefits of HiTAC technology are quantitatively demonstrated. Further on, the formation machines of NO during the HiTAC has been studied. Finally, a thermodynamic analysis basing on the 2nd law of this new combustion process will be performed in order to understand its limitation/drawbacks, the essence of the HiTAC, and the potential extension of this new combustion. 3 1.2 Thesis This thesis work will verify the fundamental phenomena of the gas combustion under high temperature and oxygen deficient atmosphere, including larger and invisible flame, and check the NOx formation and destruction mechanism. Furthermore, demonstrate the high energy efficiency, low NOx emission in a semi-industrial furnace. 1.3 Objective The generally objectives of this work are theoretical analysis and development of new combustion concepts. In particular, • Quantitative study of the flame physical properties, attention is focused on the flame volume, flame length, and flame entrainment. • NOx formation and destruction mechanisms during new combustion conditions. 1.4 Scope of the work The investigation has been performed using a single fuel jet flame with cross and co-flow during high-temperature and oxygen deficient atmosphere, and a semi-industrial furnace equipped with HiTAC burners. Experimental, numerical and theoretical analyzing investigations are adopted. Firstly, basing on the single jet flame facilities, the concepts to description of the ’invisible flame’ physical properties will be performed. The volume, length of the ‘invisible’ flame will be quantitative studied with the unique characteristic parameters of the HiTAC, i.e. high temperature and low oxygen concentrations. NO formation and destruction mechanisms during this new combustion phenomenon will be investigated. Secondly, a semi-industrial furnace equipped with HiTAC burners will be used for researching. The benefits of the HiTAC technology will be quantitatively demonstrated by mathematical models verified by experimental data. The developed concepts in above works will be used to this HiTAC application to understand the combustion in a practice furnace. 4 Finally, a thermodynamic analysis basing on the 2nd law of this new combustion process will be performed in order to understand its limitation/drawbacks, the essence of the HiTAC, and the potential extension of this new combustion, for example from air-fuel combustion to oxyfuel combustion. 5 2 State-of-the-art 2.1 History of HiTAC Weinberg [11] found the fact that preheating an ultra-lean mixture using the heat recirculation method can achieve stable combustion and expand the flammability limits of the mixture. He proposed the concept of additional enthalpy combustion, or ‘excess enthalpy combustion’, which can be defined as energy contained in the hot exhaust gas is recirculated back to the inlet air either internally or externally. Further efforts were made by British Gas and later at Hotwork International, UK [12] to use this concept for the larger scale industrial furnace, to develop both recuperative and regenerative burners. However, the problem of higher NOx emissions was met when a higher preheating air temperature is achieved. Technical breakthrough of recovering waste heat from exhaust gases using honeycomb regenerative burner for the industrial furnace was made by Mr R. Tanaka of NFK and Hasegawa T etc his cooperators [2, 13, 14, 15]. In this concept, a modern regenerative material is used to recover waste heat and the temperature of combustion air can be only 50 o C-100 oC lower than the exhaust gases temperature. They reduced NOx emission at so high preheating air temperature (over 1000 oC) using a novel idea of a low oxygen concentration. Since the combustion occurs during the condition of high-temperature and low oxygen, novel combustion phenomena appear. Since the combustion is preheating in very high level, it was first named as ‘high-preheating air combustion’, or HPAC. Late on, this was called high-temperature air combustion, i.e. HTAC or HiTAC. Further more, Gupta [15] etc gave the definition of the HiTAC as ‘the maximum temperature of the reaction zone is held constant by recirculating oxygen reduced combustion products back into the inlet air. The combustion air is preheated to temperature in excess of 1000 oC’. Another reason to use HiTAC is due to high-preheated air temperature upto 1300 oC using modern regenerative honeycomb heat exchanger to recover waste heat from exhaust gas. In this way, the utilization energy efficiency can be achieved very high level. There is another name for this high temperature combustion technology- HiTCOT. This novel combustion was also pioneering trial and industrial works carried out by Tokyo Gas with a technology of Directly Fuel Injection, called as DFI. 6 The researching of this heat recirculation technology was also conduced in European countries, for example in Germany. Wunning [16] proposed an idea of FLameless OXidation for low thermal NOx emission. In this work a FLOX was proposed, and defined as FLOX, FLameless Oxidation with high recirculation rates preheating oxidizer to a o o minimum of 650 C with combustion chamber temperatures greater than 800 C. It can be seen that the essence of this new combustion is that oxidize is diluted before combustion occurs. During this new combustion atmosphere, the color of ‘flame’ can be different, for example, green, blue, bluish depending on the dilutant used for oxidants, and degree of diluting, and different the fuels. In some certainly condition, this is no ‘visible flame’, and is referred ‘flameless’ Oxidation /combustion/ by Wunning [16], or ‘colorless combustion’ by Gupta etc [17]. MILD is another name of this technology. For example, in the work of Daily [18], he used this combustion as MILD, and defined as Moderate and Intense low Oxygen Diffusion Combustion, Uses exhaust gas recirculation to raise the oxidizer temperature and dilute the oxygen concentration to maintain a low flame temperature. Cavaliere et al [19] further studied this technology, and also explained this MILD combustion technology in the way as the temperature increase due to the reaction is lower or milder than the temperature of the preheated reactants. Diluted Combustion is another name for this technology, for example, Milani etc [20] described this technology as Diluted Combustion here, dilution means that fuel and oxidizer are mixed ‘locally’ with a ballast of inert gases before they react so that the oxygen concentration in the reactants is substantially reduced with respect to the 21% of the standard oxidizing air. Another definition would be that combustion is spread out in a large volume, which some scientists refer to as “volumetric combustion”, resulting in uniform and lower temperature of such flame. The essence of these larger flame volume and lower flame temperature is the low temperature rise during the combustion process. This is the fact that a low temperature increase makes a combustion chamber more similar as a well-stirred-reactor. 7 Again, the effect of heat release on the combustion characteristic is less at the case of a low temperature rise. For example, larger flame volume and entrainments are found during the high-temperature and oxygen deficiency condition. This concept was developed to used to the other fields, such as boiler, rotary kiln etc [21]. Blasiak etc [21] also named this combustion as a ‘quasi-isothermal combustion’ since this new combustion has a lower excess enthalpy, which firstly extends this idea to the combustion with pure oxygen. Further, innovation of this combustion phenomenon was extended to combustion with oxygen or oxygen enrichment [21]. When this new combustion is realized with air, it is generally called as ‘flameless-airfuel’. When pure oxygen or oxygen enrichment is used instead of air as the oxidizer, this is called as ‘flameless oxyfuel’. Some of study of oil [2, 22, 23] and coal [2, 23, 24] has been conduced with hightemperature air in order to achieve all the benefits obtained from gaseous fuel combustion. They are still a lot of ongoing working on this. It has been understood that high-temperature and low oxygen concentration are the unique characteristic than that of other combustion technology. The use of high temperature oxidizer has been further developed for the biomass gasification processing. In this application, referred to as High Temperature Agent Gasification (HiTAG), a highly preheated gasifying media with temperature as high as 1273 K, provides additional energy into the gasification process, which enhances the thermal decomposition of the gasified solids. HiTAG has significant advantages in the gasification of low-rank biomass fuels and traditionally unusable waste streams like for example as sludge. It can also operate efficiently on a wide range of feedstock [25, 26, 27, 28] and the HiTAG gasifier system can be built extremely compact with an atmospheric pressure, lowering component costs. Since the characteristics of the Flameless Oxidization, such as very stable combustion with low NOx levels, it has been attractive people to use this idea to high pressure combustor, for example the combustor of gas turbines. Major works in this field can be found in the 8 proceedings of Flamelesss Combustion for gas turbine, in Lund, Sweden, June 21, 2005. [29]. A lot works have been done since 1990. Here, a basic literatures related to current work will be given in next section, focusing on the study of a single jet gas-fuel and semiindustrial industries application. 2.2 Study of Single jet flame Hasegawa et al,(1997) [14] performed studying with the NKF test facility, which designed and fabricated by the Nippon Furnace Kogyo Kaisha, NFK, consists of two combustion chambers, each fitted with ceramic honeycomb regenerator, controllers for flow and switching sequence. The high temperature combustion air, produced from the regenerative test facility, and a gas jet can then be injected into this preheated oxidizer. In their work, a mixture of air, N2 and CO2 was as the oxidizer. They reported that the single jet gas flame flammability does not decrease with decreasing of oxygen concentration when the oxidizer temperature is sufficiently high. For example, the limiting temperature was 900oC for the LPG fuel. They also found that the HTAC flame had much better uniformity of temperature and the temperature fluctuations went from 197 oC under the conventional conditions down to below 5 oC for HTAC. They also reported that the flame with 3% O2 in the oxidizer and preheated to 1010 oC had very low visible radiation and a much large volume. The color of flames was reported to be blue at conventional conditions and ‘Bluish-green’ or green at HTAC firings. They found that CO2 had a stronger effect on NOx reduction then if N2 was used to dilute the oxidizer. Kishimoto et al (1997) [30] performed experiments on a similar NFK HTAC test facility. The natural gas was chosen as fuel, and N2 as dilutant and took chemiluminescence images. They found that the flame seemed to be stabilized at a large distance from the injection nozzle. When diluting the oxidizer that the dilution made the flame unstable, increasing the temperature made the flame stable again. They also performed experiments on a similar NFK HTAC test facility. The natural gas was chosen as fuel, and N2 as dilutant and took chemiluminescence images. The reported that the flame color changed from violet to blue and then to green. Their explanation was the decreasing CH emission and the increasing C2 emission 9 Amagai et al (1995)[31], made an experiment in stead-state with a electrically heat oxidizer into which they injected a co-flowing single fuel jet of propane. Both gas jet diffusion and premixed flames were checked. The results reported that both the diffusion and premixed flame were well stabilized when the temperature was over the auto-ignition temperature. This work also demonstrated that the flame length of the laminar diffusion and the premixed flames decrease with increasing temperature of the oxidizer temperature. On the other hand, the length of turbulent diffusion increases with flame-laminarization due to high temperature. Kitagawa et. al.(1998) [32] used a spectra-video camera to capture the fluctuations of LPG flame in oxidizer diluted with N2 from NFK HTAC test facility. The also measured the temperature profile in the combustion chamber with thermocouple. They calculated the vibration temperature in 2-D images of the flame using the two C2 bands. Very uniform temperature profile was found at high preheatly and low oxygen concentration, which coincided with the lowest temperature fluctuations. Uniformity and homogenization of combustion of methane and propane with highly preheated air (1000ºC) have been studied by Isiguro et al. (1998)[33]. They obtained images of OH, CH and C2 emissions for a number of experimental conditions that differed in the air preheat level and oxygen content of the air. The investigators concluded that the increase in air temperature results in a decrease of flame temperature gradients (homogenization of flames). Furthermore the higher the combustion air temperature the lower the flame fluctuations are. Plessing et al. (1998) [34] used laser-induced predissociative fluorescence and Rayleigh thermometry to examine flameless oxidation at laboratory scale. They observed that the flameless oxidation occurring in the well-stirred reactor regime. The OH concentration in the combustion zones of flameless oxidation is lower than in non-preheated undiluted turbulent premixed flames. Soon afterwards studies [35-36] have been undertook research on the effect of combustion air temperature and oxygen concentration on flame color, visibility and thermal emission spectra. Both methane and propane were used as the fuel. The spectroscopy measurements 10 checked that there were large emissions CH and C2 as well as other intermediate species for propane flame. But in the methane flame they did not measure such significant emissions. However, they noticed that there were much less intermediate species emissions for the methane flame. Bolz (1998) [36] and Gupta (1999) [37] further investigated the influences on HiTAC volume and lift-off (standoff) distances with direct flame photography. In this work, N2 was as diluted oxidizer and fuel was LPG. They found that the flame color varied from yellow, at 21 %O2 and 1100 oC, to yellow-blue flame when decreasing the oxygen concentration. And at the HiTAC firing, the flame was found to be blue-green. At 900 oC, however, the flame changed from yellow to blue, without any green color. Gupta et al, [37] further investigated found that no yellow color in flames with less than 950 oC and low oxygen concentration in the oxidizer. They concluded that the effect of oxygen concentration and oxidizer temperature on flame color be of use when designing applications which need certain radiant heat flux. Mochida and Hasegawa (2000) [38] have been developing a flame visualization technique based on the luminescence intensity ratio of C2 and CH radicals. Lille et al. (2000) [39] built an experimental facility for studying fuel jets immersing into a cross-flowing high temperature air stream and their preliminary findings have been reported. In this present, the fuel was propane and N2 was used as dilutant. The high-speed photography, direct and schlieren color visualization were used to record images of flames. They conducted that a lower oxygen concentration increases the flame size. Air temperature has an opposite effect of decreasing the flame size but not in the same proportions as an increase in oxygen content. The flame visibility decreases with decreasing oxygen concentration and increases with air temperature. The flame color changes from bright white/yellow to blue/green/yellow and is primarily influenced by oxygen concentration. Lille(2000)’s work depicted that a lower oxygen concentration increase the lift-off distance, and for high velocity fuel jet increases the lift-off distance. Lille et al (2002) [40] further investigated fuel jet combustion under the condition of high temperature and oxygen deficient air. In this work, the jet of fuel was co-axially injected into high temperature exhaust gases generated by means of a gas burner fired with Gasol 11 (mainly propane). They found that the oxygen concentration in the oxidizer have a substantial effect on flame size, luminosity, color, visibility. The flame became first bluish and then non-visible at sufficiently low oxygen concentration in the oxidizer. In the co-flow study of Fujimori et al.(1997 [41],1998 [42])) and Sato et al (1997)[43], they demonstrated that the reduction of NOx emission was largest for flames with high liftoff distance, and diluting the fuel with N2 had a considerable effect on the NOx reduction. The same results were found for air dilution. Blasiak (2001) [44] reported that addition of N2 to methane results in lower emission of NO and CO, as well as higher fuel injection velocity. Rota et al [45] presented an experiment furnace in which partially premised fuel was coaxially injected into a preheated and diluted air stream. The investigations gave that the NOx was decrease when run under HTAC conditions. Using NFK HTAC test stand, Mochida, et al.(1995) [46] investigated the flame radiation by means of CCD camera. The results showed than the relative intensity of HTAC flame is much lower, approximately one-third, and is longer than a conventional firing. They explained that the radical particles, which produce a more luminous flame, was formed in conventional firing comparing to the HTAC flame. Yuan et al.(1999) [47] investigated numerically the temperature distribution, soot formation and NO emission under the condition of the highly temperature and deficient oxygen air. They carried out the numerical model for NFK test furnace. In this work PDF model was adopted to simulate the turbulent diffusion combustion using the equilibrium chemistry method for chemical reactions. The soot formation was simulated using a two-step Tesner model and both thermal-NO and prompt NO were calculated. They shown that the effect of dilution of air stream on the decrease of maximum flame temperature various as CO2>N2>He because the difference of heat capacity and/or difference of mixing processes between fuel and diluted air. The influence of diluents on the NO emission has the same tendency as on the temperature. 12 Dong [48] performed the mathematical modeling of single jet under the highly preheated air combustion, the finite rate/eddy dissipation model and mixture fraction/ PDF are used and compared each other, general characteristics of flame were conducted. Results show that the mixture fraction/PDF model is better than the finite rate/eddy dissipation model for predicting of the thermal fields of the turbulent jet flames, and thus can give out reasonable prediction of NOx formation. Dong(2000) [49] furthermore numerical studied gas single jet under high temperature air combustion with advance flow model. In this work, the large eddy simulation (LES) and Reynold Stress Model(RSM) together with the finite rate/eddy dissipation reaction model were adopted. The results showed that these models gave a small difference in the near field predicted of flow, in contrary to the empirical constant as for example Cs in LES model which has significant influence on the predictions. Yang (2001) [50] conducted whole studying of in highly preheated and oxygen deficient air. Combustion of a single jet of propane in a cross-flowing stream of preheated and oxygen deficient air is analyzed. In the book on HiTAC (2000) [2], the characteristics of HTAC combustion numerical calculating were analyzed. HTAC firing is controlled both by chemical kinetics and by mixing. Then, the mixed fraction/PDF model and Eddy-Break-Up model are not suitable for HTAC, since the chemical reactions of both of them are assumed as infinitely fast, furthermore, PDF model was experientially obtained based on normal combustion reaction, which implicitly infer the use of ambient air. Thus, the eddy dissipation concept with multisteps reaction of fuel is an interesting suggestion for calculating HTAC at present. Furthermore, the single gas jet combustion under the condition of high temperature and deficient air was carried out. The combustion models were investigated and comparing. These models include one-step global reaction model, Jones’s four-step reaction model and Srivatsa’s four-step reaction model. The results show that Srivatsa’s model was better among the three model concerning the flame-lifted height and maximum flame temperature. Mörtberg [51] recently has used Particle Image Velocimetry-PIV to obtain information on the flow dynamics of a fuel jet injected into a crossflow of oxidizer at either a normal temperature or a very high temperature. Light emission spectroscopy was used to collect 13 information on time-averaged radical distribution in the combustion jet. Jet turbulence, time-averaged velocity distribution, fuel jet mixing, the distribution of radicals such as CH, OH, and C2 and flame photographs were investigated. The results showed delayed mixing and combustion under high-tem rapture low oxygen concentration condition. The combustion air preheats temperature and oxygen concentrations were found to have a significant effect on the burning, fuel-jet behaviour. 2.3 Study of a semi-industrial furnace with HiTAC burners In order to understand the characteristics of this novel combustion technology, HiTAC has been modeled by a single jet gas combustion in a low-oxygen partial pressure and hightemperature air environment. However progress in the use of different applications of HiTAC technology has increased the need for more information and data for furnace and process designers. In particular, it is very important to specify optimal conditions for installation of HiTAC in industrial furnaces. For these reasons, studies are performed at a larger scale where at least one set of regenerative burners systems is installed. The advantages and efficiency of HiTAC technology have been investigated and tested for semi-industrial and industrial-scale atmospheric burners by some researchers. For example, the main characteristics of flameless oxidation as promising high efficiency and low NOx combustion technology have been described by Wünning [16]. Performances of the recently developed High-cycle-Regenerative burner System- HRS have been evaluated by Suzukawa et al. [53]. Two pairs of burners were tested using a mixture of by-product gases from steel plant, with a lower heating value of 11.5 MJ/m3. The capacity of each twoburner system was 0.93 MW. The honeycomb regenerator exhibited very good performance, with preheated-air temperature being close to the furnace exhaust gas temperature, and very low levels of NOx emission recorded. The performances of two different HRSs operating on natural gas were examined by Quinqueneau et al. [54, 55]. Ranges of burner operation and levels of CO and NOx emissions were measured for both one and two-burner systems. Both systems were tested at the same capacity of 200 kW. 14 The International Flame Research Foundation-IFRF has carried out a series of semiindustrial-scale tests to identify the principal characteristics of the combustion process using high-temperature air. The fuel used in these tests included natural gas [56], oils [57] and one type of coal [58]. The behaviours of this technology were demonstrated through measuring flow field, in-furnace temperatures and heat fluxes. These experiments confirmed that HiTAC technology is an extremely attractive technology for efficient and environmentally favourable combustion. A review of above works has been undertaker by Weber [59]. Furthermore, the most important discoveries relating to ’at the burner’ heat recovery methods and NOx reduction methodology have also been summarised and reviewed by Weber (with respect to combustion air of temperature in excess of 1000oC is used) [60]. The effects that changing the size of test furnace chamber on the performance of the oneflame HRS have been investigated by Szabo et al. [61], who also performed in-flame measurements of for example O2, CO and temperature, and also studied the effects of firing rate, fuel type, NO emissions and heat transfer. Measurements of composition and heat fields have been performed inside a semi-industrial test furnace equipped with a 200kW HiTAC burner at KTH [62, 63, 64 ]. These investigations into the performances of one and two-flame burners used Liquefied-Petroleum-Gas-PLPG fuel. The results of these were also compared with a conventional high-velocity-jet burner. In Parallel with these semi-industrial tests, numerical models of the furnace have to be developed and verified. A numerical simulation in an industrial slab reheat furnace with two-flame HRS was performed by Ishii [65] and Hino [66] and respective co-researchers. These results showed that the ratio of the air-to-fuel injection velocities has a strong influence on the rate of NOx production in the furnace. However, they found that NOx models used currently might require improvement in order to properly describe NOx formation under “low-temperature” conditions. An unsteady flamelet model was employed to calculate the mild combustion mode in a laboratory-scale combustor by Coelho and fellow researchers [67]. Their results show that the steady flamelet library was unable to describe the formation of NO since this is a chemically slow process, which is sensitive to transient effect the unsteady flamelet model was able to predict the correct order of magnitude of NO emissions. 15 A steady state simulation of a one-flame burner with 200kW capacity has been performed by Pasenti et al. [68]. The numerical results showed good agreement for total radiative heat flux between measured and predicted values, but NOx emissions were substantially underpredicted. However, the trend of NOx emissions on the furnace exiting temperature was correctly represented in the predictions. IFRF tests using natural gas combustion with preheated air were modelled and reported by Dong [49]. The semi-industrial test furnace used was equipped with one burner operated under steady state conditions. The high-temperature regenerator was replaced by a precombustor, where natural gas was burned with air under lean conditions. Experiment data from this test furnace were used by Orsino et al.[69] to assess the abilities of exitedcombustion models to predict the characteristics of this new combustion technology. The combustion models tested included the eddy-break-up model, the eddy-dissipation-concept model, and PDF/mixture fraction model, had all correctly reproduced the characteristics of the high temperature air combustion with the exception of the small region located within the natural gas jet. Based on the above IFRF furnace, a recent simulation predicting the combustion characteristics with an emphasis on NOx formation and destruction of nitrogen oxides was performed by Mancini et al. [70]. They identified the location of regions of NOx formation. They concluded that the predicted values were substantially lower than the measured emissions for low preheated levels. The differences in heat transfer and combustion features between conventional highvelocity-turbulent jet flame and HiTAC flame have been numerically investigated by Yang et al [71]. The influence of stationary and permanent heat sinks on furnace heat transfer in these two different types of burner systems was also investigated and compared. The authors concluded that the HiTAC technology is more sensitive to heat sink than conventional combustion (high velocity turbulent jet flame combustion). These publications focus on the evaluation of HRS performance in relation to heat recovery efficiency, heat flux evaluation and NOx emissions. Before utilizing HiTAC in various applications, flame properties have to be known as these determine the flame stability, and the designing of the burner and the chamber. 16 Rafidi Nabil [72] has done the thermodynamic study and of the new combustion technology. His analyses display the possibilities of reducing thermodynamic irreversibility of combustion by considering an oxygen-diluted combustion process that utilizes both gasand/or heat-recirculation. Additionally, the results showed that an oxygen-diluted combustion system that utilizes oxygen as an oxidizer, in place of air, results in higher 1st and 2nd law efficiencies. Furthermore, Mathematical models of heat regenerator’s heat transfer were performed for maximized heat recovery. 2.4 Flame volume, length and entrainment It is known that the diffusion flame length can be generalized as the function of fuel jet momentum and furnace temperature by means of the Froude number, Frf [73, 74, 75]. The entrainment of the jets is the key technique solution for the industrial applications of this novel combustion technology. Jet entrainment is the radial inward flux of ambient fluid drawn into a jet. The entrainment into a turbulent jet was investigated in a number of earlier studies [76-90]. For the free nonreacting turbulent jets, Ricou and Spalding [76] gave the general, well-know expression where Ce =0.32 is the entrainment coefficient. In reacting jets, the entrainment behavior is less straightforward due to heat release and buoyancy. Important quantitative measurements and correlations of the entrainment rate based on this equation induced by turbulent gas diffusion flames are further investigated by [77-89]. For example, the constant Ce has been corrected for the cases of reacting jets by works [78, 79]. In order to get a general expression of the entrainment ratio for reacting jets, efforts have been made by works [82, 84, 85]. The influences of the heat release and/or the buoyancy force on the jet entrainment rate were investigated as well [78, 79]. 2.5 NO emissions The HiTAC flame has also proved to have many features that are superior to conventional turbulent diffusion flames. These major features include a larger flame volume, a more uniform temperature filed and a less luminous, sometimes invisible flame. NOx emissions are also kept at a very low level. This is because during the HiTAC condition, the large 17 quantities of recirculation of combustion products are entrained into the fresh reactants before combustion thus higher peak temperature is lack. As a result, thermal NOx is suppressed and much of NO maybe form mainly by mechanisms that are insignificant in most conventional combustors. A more details examination of the NO formation process in HiTAC conditions is needed in order to present a clear background to the continuously developing range of Nitrogen Oxides (NOx) control techniques and equipping HiTAC in real industrial furnaces. Although full chemical equilibriums model of NOx emission can offer more detail information, simplified chemistry models of NOx formation and destruction are more reasonably and realistic for predictions of an industrial-scale combustion chamber because of cost-efficient at this stage. A number of works for predicting the NO formation in furnaces equipped with HiTAC burners have been shown [68, 70, 71]. In these works, the existing NO models were examined. The generally conclusion can be drawn as the existing NOx models are able to predict the trend of the NOx emissions, the thermal NOx formation is predominant and the NOx reburning mechanism is of little importance in which low hydrocarbon concentration fuel is used[70]. Additionally, NO predictions are sensitive to the form of PDF used to take into account the effect of turbulence-chemistry interactions [7], and the use of the partial equilibrium or equilibrium O radical approaches to determine the O radical concentration has a strong influence on the NO predictions [2]. However, the predictions of NO emission are lower than the measured values [2, 65 66, 91, 92]. Therefore, present NOx models might require improvement to describe properly the NOx formation under ‘low-temperature’ condition [2, 65 66, 91, 92]. The N2O-intermediate mechanism is important in fuel-lean or low temperature condition. The N2O-intermediate NOx mechanism was proposed first by Malte and Pratt [93] for NO formation from molecular nitrogen (N2) via nitrous oxide (N2O). This model was further extended by some researchers [94-96]. The study [94] points out that under favourable circumstances, this mechanism may contribute to as much as 90% of the NOx formed in combustion process. The relevance of NOx formation in gas-phase from N2O has been observed indirectly, and theoretically speculated for a number of combustion systems by a number of different researchers [93, 96-99]. However, there is no any mathematical model available of reduction mechanism for the calculation of NO formation via N2O-intermediate 18 as argued in [99] that the lack of a proper model for the nitrous oxide mechanism has not allowed an exact calculation of NO concentration for HiTAC. 19 3 Methodology 3.1 Experimental faculties 3.1.1 Single Jet flame Experimental facilities The combustion of a propane gas jet in the laboratory furnace was studied. Both cross-flow and co-flow of fuel and air nozzles were considered. The schematic of the combustion chamber are shown in Figure 3.1. The combustion air for the first test case is preheated by an electrical heater, and diluted by nitrogen. The fuel and air is injected with a cross-flow arrangement as shown in Figure 3.1 (a). The combustion air for the second test case is preheated and diluted by a flue gas generator. In this case, the composition of the oxidizer is close to what can be found in a real industrial furnace. The fuel jet was injected in a co-flow arrangement to the main flow of the hot flue gases (Figure 3.1(b). The fuel studied in this work is liquefied Petroleum Gas-LPG. The components (vol%) are: CH4=0.02, C2H6=0.95, C3H8 =98.35, C4H10 =0.67. The ranges for the parameters studied are as follows: 1. Flow rate of the fuel was varied from 4.33×10-6 kg/s (0.13 nl/min) to 1.73×10-5 kg/s (0.53 nl/min), 2. Flow rate of the oxidizer was from 1.1kg/s to 1.3 kg/s, 3. Fuel preheat temperature was varied in the range between 388 K and 1073 K, 4. Air preheat temperature was in the range from 1041 K to 1573 K, 5. Oxygen concentration in the preheated air was varied from 2% to 23.2% (mass%). 20 Data logger for temperatures Upper probe Yellow color Flame zone Temperature measurements Temperature measurements Blue color Flame zone Lower probe Lift-off Distance [mm] Fuel injection Nozzle (0,5 - 0,9 mm) Thermocouple K - type Thermocouple S - type Flue gas analysis (NOx, CO, CO2, O2) Combustion gases Propane fuel (0,2 - 0,4 nlitres/min) Water cooling 0,2 0,28 0,75 0,3 Fuel Diluted and Preheated Air (a) Diluted and Preheated Air Fuel (b) Figure 3.1 Schematic of the combustion chamber for single fuel jet test furnace (a) Cross flow (b) Coflow 21 3.1.2 Classification of HiTAC burners Combustion with high-temperature preheated air diluted by hot-combustion-production has become increasingly attractive in industrial furnaces over recent years. This technology, when applied together with a modern regenerative system, offers significantly increased energy efficiency, very low CO, CO2 and NOx emissions and high quality of the product at increased production rate. For its industrial application, fuel nozzles and combustion air nozzles are arranged on the burner at a certain distance from each other, fuel and high temperature air are injected directly into the furnace at a high velocity. Because of the entrainment of these injection moments, the in-furnace gas in the zone near the burner is thoroughly mixed and its partial pressure of oxygen is lowered. The combustion stability of fuel directly injection into this zone of oxygen at low partial pressure is possible if the air preheated temperature exceeds the auto ignition temperature of the fuel. In the industrial furnace, the combustion air can be obtained a temperature of 800-1350oC, and only lowers 50oC as exhaust flue gas by recovered from exhaust flue gas improved through the very effective recovery of waste heat by a high performance heat exchanger, for example, a regenerative heat exchanger switched in the high cycle, which can recover as much as 90% of the waste heat. Thus, a large energy saving is achieved. Meantime, the temperature of fuel and air are raised well above the auto-ignition temperature of the most of gas fuel, which means that the conditions for flame stabilization are very favourable. It would be very helpful to classify the application of high- temperature air combustion (HiTAC) technology with modern regenerator burner before the study is begun. Due to the design of the HiTAC burners to control the mixing of the fuel and air jets with the furnace gases, the chemical reaction rate in these burners is lower than in conventional combustion. The typical characteristics of the HTAC technology, is its capacity to generate larger flame volumes than conventional combustion, which results in an increased heat transfer. Additionally, the more homogeneous reaction associated with HiTAC technology also implies that the heat release in the reaction zone is more widespread, leading to a more even and moderate temperature rise. Consequently, the emission of NOx can be kept low. 22 Results from previously published studies of HiTAC applications in industrial furnaces, demonstrate that high-temperature air can be obtained using a regenerative heat exchanger. In this heat exchanger, which can take the form of ceramic balls or honeycomb, the heat is periodically stored and withdrawn from the heat storage material,. Hot furnace gas and cold combustion air flows alternatively in the regenerator in contact with the regenerator material. The furnace gas transmits heat to the storage material, during the heating period. The combustion air absorbs heat from the storage material, cooling down the regenerator during the cooling period. To maintain a continuous operation at least two regenerators are required. Continuous operation is achieved in the model regenerator by switching periodically from hot, furnace gas to cold air in the regenerator with a short switching interval. It was found that a shorter switching time results in a more efficient waste-heat recovery rate and a more uniform and high temperature [52]. When more than one pair of burners are used, different firing configurations can be used to obtain a better heating performance, that is desired in special applications [5]. The switching interval used, can be varied from 4-5 seconds [52] to 60 seconds, and correspond to known applications of HiTAC, that are referred to as Highcycle Regenerative System or HRS. The two main HRS solutions currently in use feature either one or two-flame burner systems. A one-flame HRS is characterized by a single flame created by one fuel nozzle surrounded by air inlets and flue gas outlets [16, 52]. The scheme is shown in Figure 3.2. The single flame develops along the axis of the fuel-jet nozzle during the cooling and heating periods. Fuel is supplied continuously through the same nozzle. In this way a single flame can be formed with a permanent position. This position remains almost unchanged between heating and cooling periods, as the regenerators are located around the fuel jet nozzle. 23 Fresh Air F.G. F. Flue gas F.G. H.A. Figure 3.2 Scheme of one-flame HRS In a two-flame HRS, there are two separated high-cycle regenerative burners. This scheme is shown in Figure 3.3. The two burners are located in the walls of the furnace and work in pairs by a set of valves that change the direction of the air and the flue-gases according to the required switching time. Normally there are a few pairs of burners working together. Each burner has a preheated air outlet located centrally, and two fuel nozzles located laterally. When the hot furnace gas passes out through the regenerator of one of burners (heating period), the fuel nozzle of this burner is closed, and the combustion air and flame are switched off. During this time another burner operates in the combustion mode, or cooling period of the regenerator. That means the air is preheated via cooling of the regenerator, and both fuel nozzles of the fired burner are on and two flames can be created. In this way, the flame can be shifted from one burner to another in accordance with the switching time between the heating and cooling periods of the regenerator. 24 F. H.A F. F.G. Fresh Air Flue Gas Figure 3.3 Scheme of two-flame HRS F A F F A F F F.G. A F F.G. F.G. (a) F.G. F.G. F A F A F F (b) F.G. F A F (c) Figure 3.4 Two-flame regenerator system firing configuration for uniformity temperature profile in the furnace (F means Fuel, A means air and F.G. means flue gas) (a) single-side firing configuration (b) stagger firing configuration(c) counter firing configuration To discharge uniformly heated product, the furnace itself must have uniform temperature profile. The thermal profiles control available with two-flame HRS basing on composition 25 of burners with different switch time. Firing configurations that provides temperature control across the width of the furnace through three basic flame configurations (Figure 3.4) named as a) Single-side firing Mode, b) Counter Mode, and c), Stagger Mode were investigated numerical. 3.1.3 The HiTAC Test Furnace The HiTAC test used in this study is equipped with modern regenerative burner system. The outside dimensions of the furnace body are 3.5×2.2×2.2 m. Figure 3.5 represents the configuration of the HiTAC test furnace at KTH. Four tubes with an external diameter of 0.11m each and cooled with air have been installed horizontally in each corner of the furnace to remove heat from the combustion chamber. On the opposite side of the furnace to the face, there are two flue gas ducts of 0.11 m external diameter for removing hot flue gases from the furnace. The walls of the test furnace consist of two layers: an outer steel cover 5.0×10-3 m thick, and an inner layer of fibrous ceramic insulation 0.3 m thick. The inner volume of the combustion chamber is 7.2 m3. There are also a number of openings in the furnace body for measurements and observations inside the combustion chamber. 26 (a) (b) Figure 3.5 HiTAC test furnace (a)HiTAC test furnace at KTH, (b)Configuration of HiTAC test furnace with one-flame HRS. 27 22 23 11 Air-nozzle Exhaust flue nozzle D=220 350 Figure 3.6 Configuration of the HiTAC test furnace with inlet arrangements 20 140 D 3.7 92.5 (a) (b) Figure 3.7 Configuration of HiTAC test furnace with two-flame HRS (a) Two-flame HRS (b) Top view of HiTAC test furnace and test positions 28 C 11 Two Flame D 8 6 4 12 9 13 10 3 2 One Flame 1 5 A Two Flame 7 B Figure 3.8 Top view of HiTAC test furnace and test positions The furnace is designed such that two different HRS can be used. The first HRS is attached to the front of the furnace. It is a so-called one-flame system and has a thermal capacity of 200 kW.. Figure 3.5 (b) represents the computational domain of the HiTAC test furnace. Figure 3.6 gives a basically inlet configuration of an one-flame HiTAC burner The other system which is composed of two pairs of HRS is installed on the left and right sides of the furnace as shown in Figure 3.7. The 200 kW one-flame HRS with honeycomb regenerative burners was the first used in this project. The ceramic honeycomb regenerators, through which the exaust gas and combustion air are vented, are an intergral part of the burner body. Figure 3.6 shows the dimensions of the burner and locations of fuel and air injection ports. There are 12 regenerators in total, working in pairs and organised into two groups separated by intervals. 80% of flue gases are vented through the burner outlets, which is sufficient to preheat the combsution air for the desired fuel. The remainder of the exhaust gases flow out from the furance through the chimney located on the rear wall of the furnace. 29 The second calculated case assumes that the furnace is equipped with four high-cycle regenerative burners with capacity of 100 kW each. The burners (four burners marked A, B, C, D) are placed on the sidewalls of the furnace as it is shown in Figure 3.7(a). Each burner consists of one injection port for combustion air and two nozzles for fuel injection. Combustion air and fuel are injected separately. The combustion air injection port is located in the centre of the burner. The fuel nozzles are placed in the same plane on both sides of the combustion air port as shown in Figure 3.7 (b). This type of regenerator allows preheating of combustion air up to 1537 K. The fuel used in the study was LPG with a flow rate of 7.7 Nm3/h for all the studied cases. The composition of the fuel used was 0.22% CH4, 0.95% C2H6, 98.35% C3H8, and 0.67% C4H10. The air flux was around 200Nm3/s 3. 1.4 Measurement program for the semi-industrial furnace For flue gas composition in furnace measurements, a water-cooled gas-sampling probe was inserted for taking samples at various points inside the furnace. The probe was mounted at the top of the furnace on the traversing system and can be inserted down in the furnace to 1.2m from the ceiling. The movements of the probe tip in x, y and z coordinated can be carried out by a computer controlled traversing system. The probe quenches the sample and reduces the temperature down to 140 oC in a very short time and to prevent further processing of the sample. In order to avoid wash out of some species such as NO2, the sample was immediately cooled down to the below ambient temperature by means of a cooler and condensate remover. However some dissociation products, even if present in the furnace, will not be detected when using conventional cooled sampling probes since they will recombine faster than the probe can quench the reaction. 3 thermocouples were fitted with the furnace left sidewall (x=0.8m and y=−0.3m). These thermocouples are applied to measure furnace wall temperature. 30 The measurements of the flue gas parameters were made at 13 fixed locations on the horizontal plane as shown in Figure 3.8. The measurements of species were performed for 11 horizontal planes at the vertical distances with the interval 0.05m. The uncertainties associated with the probe positions are ± 10 mm in the x-and z-directions, and ± 2 mm in the y-direction. The average uncertainty of measured parameters in flue gas according to the micro-GC measurement, it is composed mainly of 98% propane, 0.9% ethane, and 0.8% butane, with a lower heating value of 93.2 MJ/Nm3. 3.2 MATHEMATICAL MODEL FOR HiTAC 3.2.1 Turbulent Combustion Model For HiTAC, the incoming preheated air is diluted with combustion products that are recirculated inside the furnace before the preheated and diluted air jet makes contact with the fuel. The temperature of the air is normally higher than the autoignition temperature of the fuel. The combustion can take place immediately after air and fuel are mixed. However, this rate of combustion is slower due to the lower partial pressure of oxygen in the combustion air. The zone of chemical reaction tends to be larger, which is quite different from conventional combustion. Studies, [2] have shown that the characteristic time of kinetics and turbulence are comparable, (Damköhle number≈1) and the two are coupled with each other. Therefore, the combustion rate is controlled by both chemical kinetics and by the turbulent-mixing. As a consequence, the combustion model that is based on the assumption that mixing–is–burned is not suitable for predicting HiTAC. To be able to accurately simulate HiTAC, when using the full reaction mechanism it is indispensable to consider all the intermediates. However, a practical simulation of an industrial furnace including a three dimensional flow with full reaction mechanism is far beyond the capability of present computers. Therefore, the most realistic solution would be to adopt a set of greatly simplified reaction mechanisms covering some intermediates. 31 Attention should be paid to the rate constant of a reaction during simulation of the HiTAC. These constants are commonly obtained for normal combustion using ambient temperature air. The same problem exists when a full reaction mechanism is used, even for the elementary reactions, since the accuracy of all the associated rate constants has not been confirmed. Therefore, the constants in models have to be optimized on the assumption that air temperature and oxygen concentration are variable. The combustion model used for HiTAC simulation must be a model capable of expressing precise reaction rates in a hightemperature and low oxygen partial pressure atmosphere. In this study, the combustion model involves both chemical-kinetic and turbulent-mixing based models. This entails evaluation of both rates locally and then taking the slower of the two as the controlling rate according to the following: RHiTAC = − min[ REBU , RKIN ] (3.1) where REBU is the corresponding turbulence-controlled rate, determined from the eddybreak-up model [101], and RKIN is the kinetic rate. Considering the used fuel in the trial whose composition is listed above, the reactions were given by: CH4+1.5O2 Æ CO+2H2O (R3.1) C2H6+2.5O2 Æ 2CO+3H2O (R3.2) C3H8+3.5O2 Æ 3CO+4H2O (R3.3) C4H10+4.5O2 Æ 4CO+5H2O (R3.4) CO +0.5O2 Æ CO2 (R3.5) The kinetically controlled reaction rate of the fuel RKIN is defined as: R KIN = CM F T β ∏ ( allj ρm j Mj v ) j e − Ea / RT (3.2) 32 Here ∏ ( x ) denotes j the product of all xj , vj are the species rate exponents, Mj the molecular weights, mj is mass fraction of the species, C, β and Ea are the pre-exponential factor, temperature exponent and activation energy for the reaction, respectively. R is the universal gas constant, T is temperature. All kinetic rates were taken from Reference [102]. In the context of the Magnussen and Hjertager model [101], the kinetic rates are deliberately set very high so that turbulent mixing is guaranteed to be the controlling rate. Mathematically, these statements translate into the following equation: R EBU = − ρε ⎡ m m ⎤ A min ⎢m F , O , B P ⎥ k sO sP ⎦ ⎣ (3.3) Where, sO = nO M O / nF M F , s P = n P M P / n F M F , A and B are empirical coefficients. In order to decide on a suitable combustion model for HiTAC, besides the eddy-break-up model, a PDF-mixture fraction model with chemical equilibrium [113]was also used. The mixture is assumed to obey the ideal gas law. The viscosity, thermal conductivity and specific heat of the mixture have been computed from the properties of individual species, and are all functions of temperature 3.2.2 NOx models The amount of NO is small and the time scale for NOx reactions is larger than the time scales for the turbulent mixing process and the combustion of hydrocarbons that control the heat-releasing reactions. Hence, therefore, it is possible to assume that the reactions involved in the NO chemistry can be decoupled from the main combustion reaction mechanism. In this work, four different mechanisms have been identified for the formation and destruction of NO, i.e.., thermal NO, prompt NO, NO reburning and N2O- intermediate mechanism. 33 3.2.2.1 Thermal NO The formation of thermal NO is determined by the following three extended Zeldovich mechanism[113]: k ,k 1b O + N 2 ←⎯1f⎯ ⎯ → NO + N ; (R3.6) ,k k 2f 2b ⎯⎯ → NO + O . N + O2 ←⎯ (R3.7) k ,k 3f 3b N + OH ←⎯ ⎯ ⎯ → NO + H (R3.8) Based on the quasi-stead state assumption for N radical concentration, the net rate of NO formation via the foregoing reaction can be determined by: d [NO ] = dt T _ NO 1 k1b [NO ] 1+ k 2 f [O2 ] + k 3 f [OH ] × (3.4) ⎡ ⎤ 2k1b (k 2b [O ][NO ]) + k 3b [H ][NO ]⎥ ⎢2k1 f [O ][N 2 ] − k 2 f [O2 ] + k 3 f [OH ] ⎢⎣ ⎥⎦ Where k i f (b ) = Ai T Bi exp(−Ci / T ) Where, T is temperature, K. The reaction constants, Ai, Bi and Ci , were taken from Baulch et al. [103]. All species concentrations used in Eq.[3.4] and (R3.6-R3.8), except for O, H and OH radical concentration, are from the main combustion simulation. In this work, the concentration of H atom is set to zero. The concentrations [O] and [OH] can be calculated from the thermodynamic equilibrium as following [104]: [O] = 3.97 × 10 5 × exp(−31090 / T ) T [O2 ]1 / 2 mol/m3 (3.5) 34 [OH ] = 1 −2 6.24 × 10 × T 0.057 × exp(8600 / T ) [O2 ]1 / 2 [H 2 O]1 / 2 mol/m3 (3.6) 3.2.2.2 Prompt-NO formation mechanism For gaseous fuels, De Soete [105] proposed a roughly estimated chemical reaction rate appropriate for the prompt-NO formation mechanism. In this rate, Missaghi et al. [106] have included a prompt factor to extend the expression to natural gas combustion. In terms of concentrations it reads, in g_mole/cm3s, E d [NO ] M 1+b b = f pr C 1+b [O2 ] [N 2 ][Fuel ]T exp(− a ) dt P _ NO RT mol/cm3s (3.7) ρ where M stands for the mixture molecular weight and ρ is the mixture density. The oxygen power b is related to oxygen mole fraction in the flame, ⎧1.0[O 2] ≤ 4.1 × 10 −3 ⎪ ⎪− 3.95 − 0.9 ln[O2 ] b=⎨ ⎪− 0.35 − 0.1ln[O2 ] ⎪0 ⎩ [O2 ] ≤ 4.1 × 10 −3 4.1 × 10 −3 ≤ [O2 ] ≤ 1.11 × 10 − 2 1.11 × 10 − 2 ≤ [O2 ] ≤ 0.03 [O2 ] ≥ 0.03 (3.8) For C3H8 the constants C and Ea take the following values : C=6.4×106 Ea= 49.65×103 s-1 cal/mole (3.9) (3.10) The prompt-factor f pr is calculated as follows [105]: 35 f pr = 4.75 + 0.0819 N c − 23.2 1 λO 2 + 32.0 1 λ 2 − 12.2 O2 1 λ3 (3.11) O2 where NC is the number of carbon atoms in the hydrocarbon and λ O 2 the local stoichiometry. 3.2.2.3 N2O-intermediate NOx The emphasis of this study in on the nitrous oxide mechanism as following [108]: N2 +O + M ← ⎯→ N 2 O + M (R3.9) N 2O + O ⎯ ⎯→ N 2 + O2 (R3.10) N 2O + H ⎯ ⎯→ N 2 + OH (R3.11) N 2 O + OH ← ⎯→ N 2 + HO2 (R3.12) N 2O + O ⎯ ⎯→ NO + NO (R3.13) N 2O + H ⎯ ⎯→ NO + NH (R3.14) The N2Oforms by reaction (R3.9) and is destroyed by reactions (R3.10) – (R3.14). Portions of this N2O are converted to NO by reactions (R3.13) and (R3.14). The kinetics rates used are listed in Table 3.1. 36 Table 3.1: Arrhenius kinetic coefficients used for nitrous oxide mechanism reactions Units:cal,mol,cm3,sec KKIN=ATbexp(-Ea/RT) Reaction R9 A b Ea 4.7e12 0 17614 Reference CECR[109] M body collision efficiencies: H2=2, O2=0.4, Ch4=2,CO2=2, CO=1.5, H2O=6,other GRI 3.0[110] species=1. R10 1.4e12 0 10810 GRI 3.0[110] R11 3.87e14 0 18880 GRI 3.0[110] R12 2.0e12 0 21060 GRI 3.0[110] R13 2.9e13 0 23150 GRI 3.0[110] R14 1.385e17 -0.5 33729 GRI 3.0[110] It is noteworthy that all reactions rates depend strongly on O, OH and H radial concentrations. Therefore, the prediction model of these radicals will be sensitive for the prediction of NO emission. The effect of turbulence on N2O formation and destruction shall be considered. The EddyDissipation-Conception model [2] has achieved some degree of success in HiTAC is. Here, it is extensively [111] used in the calculation of the chemical reaction rate of N2O-route NO model. The N2O concentration, which accounted by its production (R3.9) and its reconversion to N2 (R3.10)-(R3.12), has been calculated with this finite rate-eddy dissipation assumption. According to the model, the reaction rate of NO is given by: R EBU _ NO _ i = − ⎡ mj ⎤ Aebu min ⎢mi , ⎥ k s j ⎦⎥ ⎣⎢ ρε (3.12) where, ρ is density of mixture, kgm-3, ε is turbulence kinetic energy dissipation rate,m-2s-3, k is turbulence kinetic energy, m-2s-2. In the terminology of above, mi is mass fraction of N2 37 and N2O for (R3.9) and (R3.10)-(R3.14), respectively. mj is mass fraction of O for (R3.9), (R3.10) and (R3.13). It is mass fraction of H for (R3.11) and (R3.14). For (R3.12), it is mass fraction of OH. s j = ni M i / n j M j and n is the stoichiometric coefficient (number of moles) and M is the molecular weight. Aebu is empirical coefficients with value 4. Then, the reaction rate for reaction (R3.9)-(R3.14) entails evaluating chemical-kinetic and turbulent-mixing rates and taking the slowest as controlling, [ R N 2O _ NO _ i = − min R EBU _ NO _ i , R KIN _ i ] kg/m3s (3.13) where REBU_NO_i is the corresponding turbulence-controlled rate, determined from Equ. 3.12, and RKIN_i is the kinetics rates obtained from table 1. Furthermore, the N2O transport equation has been solved adding only reaction rates of (R3.9)--(R3.12) as chemical sources. The NO concentrations which come from (R3.13) and (R3.14) was implemented to add as a source term to the NO transport equation. The similar N2O-intermediate model has been implemented by Tobacco [111] in order to assess the effects of this NOx formation pathway with increasing pressure. 3.2.2.4 NO reburning mechanism Under fuel rich conditions, NO can be reduced by CHi radicals. The reactions involved in this global mechanism are[113]: where, k i = Ai T Bi exp(−Ci / T ) k1 NO + C x H y ⎯⎯→ HCN + ... (R3.15) k2 HCN + O2 ⎯⎯→ NO + ... (R3.16) k3 HCN + NO ⎯⎯→ N 2 + ... (R3.17) The reaction constants, Ai, Bi and Ci , are taken from Bowman [112]. The rate of depletion for NO is given by: 38 d [NO ] = k1 [CH ][NO ] + k 2 [CH 2 ][NO ] + k 3 [CH 3 ][NO ] dt (3.14) 3.2.2.5 NO source term and turbulence-chemistry interaction The NO source term due to the formation and destruction of the thermal NO, prompt NO, N2O route and NO reburning can be calculated as: ⎧ d [NO ]T _ NO d [NO ]P _ NO d [NO ]N 2O _ NO d [NO ]R _ NO ⎫ S NO = M NO ⎨ + + − ⎬ dt dt dt dt ⎩ ⎭ (3.15) The interaction between turbulent flow and chemistry is taken into account through the transport equation of NO chemical species. Thermal NO is strongly temperature dependent and so it may be argued that the variation of temperature has a great impact on NO production that species concentration fluctuation. The mean NO concentration formatted via thermal NO is computed on the basis of the single-variable probability density function (pdf) model. The Beta-function PDF is used in this study since it is widely used in turbulent combustion simulations. The mean NO concentration is calculated by solving its transport equation based on the flow field as combustion solution from the combustion simulations. ρui ∂YNO ∂Y ∂ ( ρ D NO ) + S NO = ∂xi ∂xi ∂xi (3.16) where, ρ is mean density, u is mean velocity in the i direction, and YNO is the mean mass fraction of NO. The mean source term, S NO , is determined from above NO mechanisms. 39 3.2.3 Other Models Used The flows in the industrial furnace are turbulent. It follows that the performance predictions of combustors depend very much on the turbulence model adopted. With the requirement of a million nodes per cubic millimeter, it is clear that the application of DNS (direct numerical simulation) for engineering purposes is not practical for application here. LES (Large eddy simulation) seems to have a bright future, but more research on LES is required. Dong [49] has carried out a simulation of a single fuel jet flow in hightemperature diluted air combustion. It was found that advanced turbulent models, such as LES and RSM, gave small differences in the near field when predicting the flow. However, the empirical constants, for example Cs in the LES model, have a significant influence on the predictions. This implies that the empirical constants in traditional models must be adjustable to be able to obtain the best performance for HiTAC simulations. The k-ε model remains the obvious starting point, especially for diffusion flames for engineering calculations. It has been verified that it is robust and efficient for most engineering calculation purposes in the range it can be used. For flows in HiTAC, due to the larger reaction zone and the similarities to a ‘well-stirred reactor’, the assumption of non isotropy of the turbulence is weak. Furthermore, the buoyancy in the furnace is relatively small compared to conventional combustion due to both the preheated air and lack of a distinct flame zone. Consequently the k-ε model is chosen to be used in this paper. Additionally, radiation was handled using the discrete transfer method [113]. Radiation properties of flue gas were assumed to be of the grey body type and were temperature- or concentration-dependent, or both. Available absorption coefficient model called the weighted-sum-of-grey-gases model or WSGG was used to determine the absorption coefficient. 3.3 DEVELOPED CONCEPTS FOR DESCRIBING HiTAC CHARACTERISTICS 40 HiTAC has many characteristics that are completely different from conventional combustion. For example, the HiTAC flame is less visible than the flame from conventional combustion where there is a high concentration of oxygen by volume (more than or equal to 21%, i.e air). Therefore it is generally accepted, that flame length is not a suitable parameter for characterizing flame size for HiTAC. Instead, it is necessary to characterize the flame shape and size using a comprehensive numerical simulation. To describe the flame under HiTAC conditions, the oxidation mixture ratio is used in this work. The oxidation mixture ratio allows the combustion progress to be estimated and to be calculated as the mass fraction of oxygen divided by the mass fraction of oxygen plus the amount of oxygen needed to achieve complete combustion at any point in the combustion chamber, as follows: mO mO + ∑ s c m F , c Ro = c where s = nO M O / nF M F (3.17) . This ratio has a value of Ro =1 at the air inlet or when the combustion is completed, and a value of Ro = 0 at the fuel inlet. To determine flame border through this parameter, the critical Rcr must be given. Eq. 3.17 can be transferred as following: Ro = 1 1 + ∑ sc c m F ,c (3.18) mO For HiTAC technology, the preheated air temperature is above the fuel’s autoignition temperature, therefore, the fuel flammability limit plays the main role for combustion stability. Thus, the criteria Rcr can be determined through the flue flammability limit φ in Eq. 3.19. Rcr = 1 1 + ∑ scφc (3.19) c 41 The lean flammability limits for different fuel species has been used to indicate the outside border of the flame, and the rich flammability limits of them are used to given the insider border of the flame. According to Glassman [114], the flammability lean limit of propane in air and oxygen is 2 % (fuel volume percent), and this value of CO in air is 12%, then RO= 0.99 is assumed to indicate a flame border. Thus the flame volume can be approximate defined when 0 ≤ RO ≤ 0.99 (3.20) In order to describe the overall radiation field of a flame, the radiant fraction (frad) is used. It is defined as the ratio of net radiative heat loss from the flame (Qrad) to the total heat released during combustion (QF) as follows: f rad = Qrad QF (3.21) To evaluate gas temperature field uniformity inside the furnace, a furnace-gas-temperature uniformity-ratio, Rtu, is defined as follows: ⎛ (T − T ) ⎞ ⎟⎟ = ∑ ⎜⎜ i Rtu ⎝ T ⎠ 2 (3.22) where Ti [K] is the temperature of calculated cell number and T [K] is the average temperature in the furnace. When Rtu =0, there is no gas temperature gradient inside the furnace. For HiTAC flames, to be able to characterize the flame volume in relation to the volume of the combustion chamber, a dimensionless coefficient called Furnace Flame Occupation Coefficient (FOC), RFOC is defined as the ratio between the flame and furnace volume: RFOC = Vf VF (3.23) 42 where, Vf [m3] is flame volume calculated according to the relationship in equation 5, and VF [m3] is the furnace volume calculated from the geometrical dimensions of the furnace. Combustion intensity is also a very important parameter for designing the furnace and the burner. To evaluate quantitatively the chemical reaction intensity in the furnace and especially in the chemical reaction zone (flame), two parameters are used. One is Flame Heat Release (FHR), which is defined as the ratio of heat released inside the flame zone (Qf ) to the flame volume (Vf): q FHR = Qf Vf (3.25) where Qf is obtained as follows: Qf = ∫ Vf ∑ q dV c (3.26) Here qc (kW/m3) is local heat release of different fuel species. Another parameter used in this work is the ratio between the heat released by the flame zone (Qf) and total heat released inside the combustion chamber, (QF). It is named Flame Heat Occupation Coefficient (FHOC), and defined as follows: RFHOC = Qf QF (3.27) where, QF is calculated according to: QF = ∫ VF ∑ q dV c (3.28) Furthermore the entrainment ratio of the nozzle is a good parameter for the description of the internal recirculation of the flue gas which plays an important role in the HRS systems. 43 Due to the interaction between fuel and air nozzles for one-flame HRS, the entrainment of a single nozzle is not a suitable parameter for the characterization of the internal flue gas recirculation. Therefore, the entrainment ratio must include the interactions between the nozzles. In order to describe the interactions between the fuel and air nozzles, flame entrainment ratio is more efficient. The flame entrainment ratio (Rfe) is defined as the following: R fe = mf m0 (3.29) Here, m0 (kg/s) and mf (kg/s) represent the initial total mass flow rates and mass flow rates through the cross section of the flame respectively. 44 4 Cross-flow single jet flame study Combustion of a single fuel jet of propane in a laboratory furnace was studied numerically. Furnace’s combustion chamber has dimension of 0.16 × 0.2 × 0.28 m. Computational domain and meshes of HiTAC test furnace can be seen in figure 2. Fuel nozzle is placed on the wall in a cross-flow to the main flow of oxygen deficient and preheated air. Flow rates of preheated air and fuel were kept constant and equal to 3.333 × 10-3 m3s-1 and 5.0 × 10-6 m3s-1, respectively. Fuel preheat temperature was in the range from 288 K to 873 K and air preheat temperature from 1041 K to 1273 K. Oxygen concentration in the preheated air varied from 2% (mass) to 18%( mass). Reynolds numbers were 1162 and 3800 for the preheated air and fuel gas inlet respectively. The blowing ratio as defined in the paper by Hasselbrink et. al. [117] is 160. At this stage of the study the thermal decomposition of propane at elevated temperature before outlet from the gas nozzle was not considered. Adiabatic wall boundary conditions are assumed in the heat transfer model (i.e. zero heat flux). 4.1 Study of mathematical modelling The computational domain, representing laboratory test furnace, was divided into 23148 computational cells using unstructured meshes with embedded refinement to refine the fuel nozzle. Figure 4.1 displays the spatial discretization of the computation domain. 45 Local refined mesh near the fuel jet Fig.4.1 3D-computational domain and meshes of HTAC test furnace Gas temperature profiles and flame zones are shown in Figures 4.2 and 4.3. Figure 4.2 shows that the maximum gas temperature predicted with use of PDF combustion model is equal to 1952 K but it is equal to 1846 K if the EBU combustion model is employed. Also areas of maximum gas temperature within the flame are predicted larger when the PDF model was used. The lower peak temperature seems more realistic when oxygen concentration is only 10%. Moreover, one can see from Figure 8 that the flame volume as defined by means of the oxidation mixture ratio was predicted smaller when the PDF model is used. Also it can be noticed that the flame size and shape predicted by the EBU model is visually similar to the flame presented at photography (Figure 4.3c) taken. Based on these facts, it was assumed that the HiTAC flame predictions with the EBU model are more realistic. Thus, the EBU model was further studied as more applicable to HiTAC flame modelling. 46 (a) pdf (b) ebu Fig.4.2 Predicted temperature profiles for 10% oxygen in the air preheated upto 1041K and at fuel inlet temperature equal to 473 K. (a) PDF model; (b)EBU model (a) (b) (c) (a) PDF model; (b)EBU model, (c) flame photograph Fig.4.3 Predicted distributions of oxidation mixture ratio for 10% oxygen in the air preheated upto 1041K and at fuel inlet temperature equal to 473 K. Two empirical coefficients, A and B, are involved in the rate of reaction equation (3.3) in the EBU model. It has often proved to be necessary to adjust the A and B to obtain the good performance for a particular application. B is taken into account to inhibit reaction where the temperature is low. However in HiTAC combustion, the air temperature is higher than 47 fuel self-ignition temperature, the influence of empirical B is insignificant, so B is set as normal constant 0.5. Concerning of A, it is known that the fuel consumption decreases when A decreases according to reaction equation as the reaction rate will also decreases. Many experimental works proved that HiTAC combustion rate is slower than traditional combustion. This indicates that A should be less than the nominal value 4. Thus, simulation for the following three different values of A that is 4.0, 2.0 and 1.0 was performed. Figure 4.4 shows that predicted temperature field is influenced by value of the A coefficient. Lower value of the A coefficient gives smaller temperature gradients in the flame. Clearly it is seen that area of flame with the highest temperature is reduced when the lower value of the A coefficient is applied. (a (b) (c) Fig. 4.4 Predicted gas temperature profiles for 10% oxygen in the air preheated upto 1041K and at fuel inlet temperature equal to 288 K. (a) A=4.0, (b) A=2.0, (c) A=1.0 Influence of the A value on the oxidation mixture ratio is shown in Figure 4.5. Numerical predictions demonstrate that the flame volume increases with the reduction in A value. It shows also more uniform distribution of the fuel inside the flame when smaller values of the A are used. It confirms also that the combustion rate is slow according to equation (3.3). 48 (a) (b) (c) Fig. 4.5 Predicted distributions of the oxidation mixture ratio for 10% oxygen in the air preheated upto 1041 K and at fuel inlet temperature equal to 288 K. (a) A = 4.0, (b) A = 2.0, (c) A = 1.0 4.2 Flame Volume Ratio, Rf Consequently the EBU model was used to study influence of the fuel preheat temperature on the combustion. Figure 4.6 shows that the flame volume ratio Rf depends very much on oxygen concentration in the preheated combustion air, temperature of the combustion air and the fuel inlet temperature. Flame volume ratio always increases when the oxygen concentration in the preheated air is reduced. This is clearer if oxygen concentration in the preheated air is below 5%. 49 250 225 200 175 Rf 150 125 100 75 50 25 0 0 2 4 6 8 10 12 Oxygen Concentration (mass%) Ta=1041,TF=288 Ta=1041,TF=573 Ta=1173,TF=573 Ta=1273,TF=288 Ta=1273,TF=573 14 16 18 20 Ta=1041,TF=473 Ta=1173,TF=288 Ta=1173,TF=873 Ta=1273,TF=473 Ta=1273,TF=873 Fig.4.6 Rf versus oxygen concentration for various air and fuel temperatures (K) Elevating fuel temperature leads to reduction of the flame volume ratio Rf at constant oxygen concentration. For example when oxygen concentration is equal to 5% and combustion air temperature is equal to 1273 K, the Rf in case of the fuel inlet temperature equal to 288K is 6 times larger than in case of the fuel inlet temperature equal to 873 K. At 10% of oxygen concentration, this relation is 5.3, and at 18% of oxygen concentration, it is equal to 8.7. Reason for this is decrease of the fuel density with increase temperature thus increase of fuel inlet velocity at constant fuel fluxes. Changes of density and fuel inlet velocity are proportional to the fuel temperature. Increasing of initial velocity of the fuel jet improves the mixing between fuel and preheated air what results in flame volume decreasing. Improvement of mixing is approximately proportional to the fuel inlet velocity. Therefore, reduction of Rf is also proportional to the fuel inlet temperature. Figure 4.7 also shows that combustion air temperature has much less significant influence on the flame volume at constant oxygen concentration and fuel temperature. For the investigated temperature range (1041 K–1273 K) of the preheated air flame volume was found almost constant at fixed oxygen concentration and fuel inlet temperature. 50 120 100 Rf 80 60 40 20 0 1030 1080 1130 1180 Preheated air temperature (K) Oxyg.=5%,TF=288 Oxyg.=10%,TF=288 Oxyg.=10%,TF=473 Oxyg.=5%,TF=573 1230 1280 Oxyg.=10%,TF=573 Oxyg.=18%,TF=573 Oxyg.=5%,TF=873 Fig.4.7 Rf versus preheated air temperature for oxygen concentration and fuel temperatures (K) Generally it can be concluded that the largest flames are obtained at the lowest oxygen concentration in the combustion air and at lowest fuel and preheated air temperatures. Assuming the combustion air temperature equal to 1273 K and the fuel inlet temperature equal to 473K the HiTAC flame at oxygen concentration equal to 5 % is more then eight times larger than the flame at oxygen concentration equal to 18%. Flame volume in case of 2% oxygen concentration and 288 K fuel temperature is 223 times larger than that in case of 18% oxygen concentration and 873 K fuel temperature. The performed numerical experiments show that HiTAC process is spread over much larger volume then conventional turbulent diffusion flame. It confirms that the HiTAC is a large volume combustion with reduced combustion rate. From the above, it can be also stated that three primary factors determine flame volume. These are as follows: 1 - initial momentum of the fuel jet; 2 - oxygen concentration in combustion air; 3 - ratio of fuel inlet density to ambient gas density, ρ F ρ 0 . 51 4.3 Mean Residence Time and Flame Peak Temperature Figure 4.8 shows that the mean residence time, τ R increases with decrease of oxygen concentration. This dependence is almost consistent with change of flame volume ratio, Rf as showed in Figure 4.6. It confirms primary importance of oxygen concentration and fuel temperature on mean residence time. Fuel temperature influences the value of τ R , but not as much as the molar fraction of oxygen. For example, for air temperature equal to 1273K and fuel temperature equal to 473K, the mean residence time at the case of 5% oxygen concentration is 6.7 times larger then at the case of 18%. When oxygen concentration is 5% and combustion air temperature is 1273 K, the mean residence time in case of fuel temperature equal to 288K is 2.9 times larger than that in case of fuel temperature equal to 873 K. These values are somewhat less than changes of Rf at the same conditions as the flame density compensates. It can be concluded that the mean residence time increases with reduction of oxygen concentration as well as with decrease of fuel inlet temperature. 0,14 0,12 0,1 0,08 0,06 0,04 0,02 0 0 2 4 6 8 10 12 14 16 18 20 Oxygen Concentration (mass%) Ta=1041,TF=288 Ta=1173,TF=573 Ta=1273,TF=473 Ta=1273,TF=873 Ta=1173,TF=288 Ta=1173,TF=873 Ta=1273,TF=573 Fig.4.8 Mean residence time (s) versus oxygen concentration and combustion air and fuel temperature (K) 52 The mean residence time also slightly increases with decrease of the preheated air temperature. For example, assuming oxygen content equal to 5% and fuel temperature equal to 473K, the changes of local residence time is 1.12 times for combustion air temperature from 1273K down to 1041K. This is because flame density increases with reduction of combustion air temperature. It is obvious as showed in Figure 4.9 that for any preheated air and fuel temperature, flame peak temperature approaches maximum for the highest oxygen concentration, and minimum for lowest oxygen concentration. At any preheated air temperature the peak temperatures fall slightly for constant oxygen concentration and at reduced fuel inlet temperature. 2500 Tmax , K 2300 2100 1900 1700 1500 0 2 4 6 8 10 12 14 16 18 20 Oxygen Concentration (mass%) Ta=1173,TF=288 Ta=1173,TF=873 Ta=1273,TF=473 Ta=1173,TF=573 Ta=1273,TF=288 Ta=1273,TF=873 Fig.4.9 Peak flame temperature (Tmax) versus oxygen concentration for various combustion air and fuel temperature(K) 4.4 Gas Temperature Uniformity Ratio, Rtu 53 Gas temperature uniformity ratio, Rtu versus oxygen concentration for various air and fuel temperatures is shown in Fig.4.10. The Rtu decreases with reduction of oxygen concentration, which indicates the more uniform temperature field at lower oxygen content. Rtu in the case of oxygen concentration equal to 5 % and at fuel inlet temperature equal to 288 K is two times lower than in the case of oxygen concentration equal to 18%. Increase of temperature field uniformity results from the reduction of flame peak temperature and from increasing of the flame volume Rf at lower oxygen concentrations. 40 36 Rtu 32 28 24 20 16 12 0 2 4 6 8 10 12 14 Oxygen Concentration (mass%) Ta=1041,TF=288 Ta=1041,TF=578 Ta=1273,TF=473 Ta=1273,TF=873 16 18 20 Ta=1041,TF=473 Ta=1273,TF=288 Ta=1273,TF=573 Ta=1173,TF=288 Fig.4.10 Temperature uniformity ratio, Rtu versus oxygen concentration for various air and fuel temperatures(K) Influence of the fuel inlet temperature on the temperature uniformity ratio Rtu is difficult to detect. In general it was noticed that at lower oxygen concentration the influence of fuel inlet temperature is more distinct. At high oxygen levels, for example at 18% oxygen concentration, the magnitudes of Rtu at different fuel and air temperatures are very close to each other. It is apparent that the average temperature T and the temperature of ith computational cell, Ti in test furnace increase when the fuel temperature increases. 54 4.5 Conclusions 1) Concept of the flame volume ratio, Rf introduced in this work was used to describe the flame volume changes during High Temperature Air Combustion. 2) It was showed that High Temperature Air Combustion is spread over much larger volume then conventional combustion. 3) Flame volume was found almost constant for the investigated temperature range (1041 K–1273 K) of the preheated air at fixed oxygen concentration and fuel inlet temperature 4) Concept of gas temperature uniformity ratio Rtu introduced in this work was used to characterise the temperature uniformity changes inside the flame. 5) Mean residence time of fuel gas parcels inside the flame volume increases with reduction of oxygen concentration as well as with decrease of fuel inlet temperature and slightly increases with decrease of the preheated air temperature. 6) Increase of the fuel inlet temperature results in smaller flame, shorter mean residence time, smaller temperature peaks and lower formation of NO. 55 5 Flame Length and volume study in co-flow single jet flame The studies of global flame features showed the flame color to change from yellow to blue to bluish-green to green, and in some cases hybrid and purple color flame was also observed. Under certain conditions flameless or colorless oxidation of the fuel has also been demonstrated. It is generally accepted that HiTAC flame length is not enough for characterizing flame properties of high temperature air combustion because of its poor visibility and big chemical reaction volume. Quantitative numerical studies were performed in work, where the HiTAC flame volume was defined to demonstrate the HiTAC flame properties and its changes depending on process parameters were studied. It is known that the diffusion flame length can be generalized as the function of fuel jet momentum and furnace temperature by means of the Froude number, Frf [116]. However, there is little knowledge available about the influences of high temperature combustion air and low oxygen concentration on the flame length. It has proved that the temperature profile in combustion chamber is more uniform in the HiTAC condition although the combustion air is preheated to very high temperature. This implies that the buoyant force in furnace is smaller. Consequently, the Frf is larger than that in the case of conventional combustion with the same initial jet momentum. Therefore, the constants of the flame Frf used to assess momentum– or buoyancy–control have to be optimized on the assumption that air temperature and oxygen concentration are changeable. The present study intends to fill this gap. The “flame” is determined in this paper according to the flammability limits of the combustible gases in the combustion chamber during the HiTAC condition. The HiTAC flame length and volume in co-axial flow with high-temperature and low-oxygen concentration oxidizer were numerically studied. The “flame” length and volume describe the physical properties of the chemical reaction zone in the combustion chamber. Consequently, they are referred to as a “chemical” flame length or height [117] and volume, to distinguish them from luminous flame lengths and volumes determined based on visual observations. The flame length and volume in the latter part of this work mean chemical flame length and volume. The HiTAC “flame” length and volume in coaxial flow with a high temperature and low-oxygen-concentration oxidizer were studied. The studied 56 parameters included oxygen concentration in oxidizer, fuel and oxidizer temperatures, firing rate, and diameter of fuel nozzle. The HiTAC flame volume was also expected to be synthesized as a function of these various parameters. 5.1 Flame appearance Gas jet flame co-axial with flow of flue gases also at various oxygen molar fractions in the oxidizer is shown in Figure 5.1. Combustion of fuel gas in hot and oxygen deficient flue gases appeared to be very stable and complete even at very low oxygen concentration. (a) 10% O2 (b) 12.8% O2 (c) 13.3% O2 (d) 16.8% O2 (e) 21% O2 Figure 5.1 Flame appearance for the LPG jet co-flowing vertically with hot and oxygen depleted flue gas. Fuel nozzle diameter 0.5 mm. Fuel jet velocity equals to 25.5 m/s. Flue gas velocity equals to 0.98 m/s [40]. This experiment also shows that reduced oxygen concentration increases the flame size, lift-off distance and decreases luminosity and visibility. Flame becomes first bluish and next non-visible. It is important to stress that the substantial differences were noticed particularly in phenomena of heat transfer by radiation. Flame formed by burning fuel gas jet in low oxygen content flue gases has higher emissivity what results also in much higher total radiative heat flux. 5.2 Effect of oxygen concentration 57 Gas jet flame co-axial with flow of flue gases also at various oxygen molar fractions in the oxidizer is shown in Figure 5.2. In this figure, the flame pictures were taken by a 60×60 mm Hasselblad camera through the window opened in the front wall of the combustion chamber. This experiment also showed that reduced oxygen concentration increases the flame size, lift-off distance and decreases luminosity and visibility. Flame becomes first bluish and next non-visible. Experimental flame lengths were obtained by visual determination as shown in Figure 5.2. For the cases of [O2] equal to 16.8, 12.8 and 11.0, they were 0.25 m, 0.30 m and 0.33 m, respectively. (b) O2=12.8 % Length Length Length (a) O2=11.0% (c) O2=16.8% Figure 5.2 Flame apparence versus oxygen concentration To = 1173 K, dF = 5.E-4 m, QF = 0.01 g/s, TF =299K Figure 5.3 shows the flame shape and size defined according to Eq. 1. Consequently calculated flame length and volume of LPG flames were showed in Figure 19. Comparison with the experimental results, predicted values are larger. This is because of the difference of definitions for both predicted and measured. However, the changing trend was similar and the agreement can be accepted. 58 (a) 5% O2 (b) 10% O2 (c) 12.8% O2 (d) 15.8% O2 (e) 21% O2 RO Figure 5.3 Predicted flame shape and size for different oxygen concentration (To = 1173 K, dF = 5.E-4 m, QF = 0.01 g/s, TF =299K) From Figure 5.4 one can further see that drop in oxygen concentration in the oxidizer increases the flame length as well as the flame volume. The influence of oxygen concentration on flame volume is more visible. When the oxygen concentration varies from 21% down to 5%, the flame volume increases 13.2 times, but the flame is barely 2.4 times longer. It is obvious because the volume is proportional to the cubic of the length if the shape is similar. The rising trend in flame volume is clear if the oxygen concentration is below 10%. The possibly explanation is that although the chemical reaction between fuel and oxidizer is very fast due to the high oxidizer temperature, because of deficiency of oxygen concentration in the combustion regime, the combustion can not complete in small volume and the flame needs to disperse in order to complete oxidizing reaction of hydrocarbon until the combustibles are consumed. Therefore, this combustion phenomenon is very similar as Well Stirred Reactor (WSR). These behaviours offer further quantitative evidence for ‘volumetric combustion’ instead of ‘flame front combustion’ for HiTAC. 59 1,0 1,2E-03 0,8 Experimental Length 9,0E-04 Flame Volume 0,6 y = 0,1072x -0,5674 0,4 6,0E-04 3,0E-04 0,2 Flame Volume, m3 Flame Length, m Flame Length y = 3E-06x -1,7052 0,0 0,0E+00 0 0,05 0,1 0,15 0,2 0,25 O2 content in oxidizer(vol%) Figure 5.4 Length and volume of HiTAC flame versus oxygen concentration. (To =1173 K. dF = 5.E-4 m. QF = 0.01 g/s. TF =299K) Figure 5.5 gives the temperature distribution on the central vertical cross section versus the oxygen concentration in the oxidizer. It is obvious that the flame peak temperature approaches maximum for the highest oxygen concentration, and minimum for lowest oxygen concentration. This implies that temperature distribution tends to be more uniform as the oxygen concentration decreases. The heat release in such conditions is also distributed, leading to a dispersed and moderate temperature rise. It can be concluded that the lower the oxygen concentration in the oxidizer, the bigger the reaction zone. To see the relationship between oxygen concentration and the maximum temperature in the furnace, these values were plotted in Figure 5.6. The peak temperature decreases linearly with the reduction of oxygen concentration. 60 (a) 5% Oxy. (b) 10% Oxy. (c) 12.8% Oxy. (d) 15.8% Oxy. (e) 21% Oxy. T, K Figure 5.5 Predicted temperature profiles for different oxygen concentration. (To =1173 K. dF = 5.E-4 m. QF = 0.01 g/s. TF =299K) Changes of the peak temperature with change of oxidizer temperature were shown in Figure 5.7. However, these changes are weak in comparison to its changes with oxygen content in the oxidizer. 61 Flame maximum temperature, K 3000 Tair=1173K 2500 2000 1500 1000 0 0,05 0,1 0,15 0,2 Oxygen concentration, vol% 0,25 Figure 5.6 Flame maximum temperatures versus oxygen concentration. ( To =1173K. dF = 5.E-4 m. QF = 0.01 g/s. TF =299K) Flame maximum temperature, K 3000 Oxygen=12.8(vol%) 2500 2000 y = 0,5892x + 1312,8 1500 1000 1000 1200 1400 1600 Preheated air temperature, K 1800 Figure 5.7 Flame maximum temperatures versus air temperature. ( [O2]=12.8%. dF = 5.E-4 m. QF = 0.01 g/s. TF =299K) 62 These results were consistent with experimental results presented in work [15, 17, 100] and theoretical analysis results in works [2]. Although in these cases, fuel jet in a cross-flow of oxidizer was investigated. Maximum temperature of flame, (Tf,max) of common fuel during preheated and diluted air combustion can be estimated using the following relationship: T f ,max = To + Tad 21 /[O2 ] + 1 (5.1) 5.3 Effect of oxidizer temperature Figure 5.8 shows influences of oxidizer temperature on the HiTAC flame length and volume at 12.8% of oxygen concentration. The Reynolds number of the fuel jet was 3316. The results show that the flame length is only slightly higher for higher temperature of oxidizer. This is because, as the inlet temperature is increased, mass flux by turbulent mixing is reduced due to decrease of air density. Meanwhile, the small expansion of the burnt gas does not suppress the turbulent mixing. These effects are considered to compensate each other. The same results have been experimentally observed by Fujimori and co-workers [41]. The results also show that increase in preheated temperature of oxidizer increases the flame volume by about 40%. This is because the low air density at high temperature causes less diffusion of oxidant into the fuel flow. Less penetration of air into the fuel jet results in flame volume increase. 63 3,0E-04 y = 2E-07x - 9E-06 0,8 2,4E-04 0,6 1,8E-04 y = 1E-04x + 0,2352 0,4 1,2E-04 Flame Length 0,2 6,0E-05 Flame Volume, m3 Flame Length, m 1,0 Flame Volume 0,0 0,0E+00 1000 1200 1400 1600 1800 Air preheated temperature, K Figure 5.8 Length and volume of the HiTAC flame versus oxidizer temperature. ([O2]=12.8%. dF = 5.E-4 m. QF = 0.01 g/s. TF =299K) 5.4 Effect of fuel temperature Figure 5.9 shows influence of the fuel temperature on the HiTAC flame length and volume at 10% of oxygen concentration. In these cases, in order to simplify the numerical calculations, the pyrolysis of fuel has not been taken into account. It can be seen that increase of fuel temperature leads to decrease of both flame length and volume. Flame volume in the case of 299 K fuel temperature is 2.9 times larger than that in the case of 1073 K. The reason for this is that decrease of the fuel density results in increase of fuel inlet velocity at constant fuel flow rate, and changes in density and fuel inlet velocity are proportional to the fuel temperature. Increasing initial velocity of the fuel jet improves the mixing between fuel and oxidizer, which results in flame volume decrease. 64 1,0 4,0E-04 Flame Length, m Flame Volume 3,0E-04 y = 3E-06x -1,7052 2,0E-04 0,6 0,4 0,2 1,0E-04 y = -0,0002x + 0,441 0,0 Flame Volume, m3 Flame Length 0,8 0,0E+00 273 473 673 873 1073 Fuel preheated temperature, K Figure 5.9 Length and volume of the HiTAC flame versus fuel temperature. [O2]=10%. dF = 5.E-4 m. QF = 0.01 g/s. TO =1173K For flame length, although the trend is similar, the effect of fuel temperature is not so pronounced. Only 1.45 times the flame length is increased. The shape and size of flame at different fuel temperatures were shown in Figure 5.10. It can be seen that the changes in flame volume comes from not only flame’s length, but also flame’s diameter. This can explain the above behaviours. The function of the flame length to the fuel temperature is linear, and the flame volume is power function of the fuel temperature. 65 QF =0.01 g/s QF =0.01 g/s QF =0.04 g/s TF=299 K TF=1073 K TF=299 K (a) (b) RO (c) Figure 5.10 Length and volume of the HiTAC flame versus fuel firing rate. (TO =1173K. dF = 5.E-4 m.) 5.5 Effect of fuel flow rates Figure 5.11 depicts the influence of fuel flow rate on flame length and volume versus oxygen concentration. The fuel mass flow rates are 0.01, 0.02 and 0.04 g/s respectively. Flame length was found to be nearly constant for each oxygen concentration and is almost independent of the fuel flow rate. This may be because the studied cases were in the range of turbulent flame, the Reynolds numbers of fuel nozzle were 3316, 6631 and 13263, respectively. The flame volume increases with increase in fuel flow rate. The shape and size of flame at different fuel flow rates are given (a) and (c) in Figure 5.11. It is obvious that the diameter of the flame increases with the increase of fuel flow rate. The increasing trend is higher at lower oxygen concentration. For example when oxygen concentration is equal to 5%, flame volume in the case of fuel flow rate equal to 0.8 kW is 2.3 times larger than in the case of 66 the fuel flow rate equal to 0.2 kW. At 10% oxygen concentration, this relationship is 1.8, and at 15.8% oxygen concentration, it is equal to 1.6. This is the fact that the mean residence time increases with either decrease of the oxygen concentration, or increase of 0,9 3,E-03 0,6 2,E-03 0,3 1,E-03 0,0 0,E+00 1,0 0,0 0,2 0,4 0,6 0,8 Volum e, m 3 Length, m fuel flow rate. Fuel firing rate, kW L. at 15.8%[O2] L. at 10%[O2] L.at 5%[O2] V. at 15.8%[O2] V. at 10%[O2] V.at 5%[O2] Figure 5.11 Length and volume of the HiTAC flame at different cases ( [O2]=10%. dF = 5.E-4 m. TO =1173K) 5.6 Effect of fuel nozzle diameter The influences of diameter of fuel nozzle on the HiTAC flame length and volume at the constant fuel flow rate were showed in Figure 5.12. The fuel velocity at the fuel nozzle outlet was kept constant. The increases of the fuel nozzle diameter were obtained by elevated fuel temperature. The influence of the fuel nozzle diameter on flame length and 67 volume can be negligible for the studied cases. This is because the changes of mixing between fuel and the oxidizer are small for the studied cases because of the constant velocities of fuel and air. It can be argued that the combustion with high temperature and low oxygen concentration oxidizer is diffusion combustion although it is ‘volumetric combustion’ instead of ‘flame front’ combustion. 4,E-04 3,E-04 0,6 2,E-04 0,3 Length Volume 0,0 1,E-04 0,E+00 Fla m e Vo lum e, m 3 Flam e Length, m 0,9 2,E-04 4,E-04 6,E-04 8,E-04 1,E-03 Diameter of fuel nozzle, m Figure 5.12 HiTAC Flame length and volume versus diameter of fuel nozzle. ([O2]=10%. TO =1173K. TF=299K QF =0.01 g/s) 5.7 Scaling analysis Generally, Froude number is defined to establish momentum-controlled and buoyancycontrolled regimes. For turbulent jet diffusion flames, the following definition of flame Froude number is used [73]: 68 Fr f = v F f s3 / 2 ⎡ ΔT f ⎤ ρ ( F )1 / 4 ⎢ gd F ⎥ ρ∞ ⎣ T∞ ⎦ 1/ 2 (5.2) where ΔT f was estimated using ΔT f = T f ,max − To , and fs = 1 (5.3) (mo / m F ) stoic + 1 Flame length and volume were plotted against the flame Froude number, Frf, as the abscissa and a dimensionless flame length L* and V* as the ordinates for flame length and flame volume, respectively in Figure 5.13. L* = [d Lf fs F ( ρ F / ρ o )1 / 2 ] (5.4) and V* = [ V f f s2 L f × d F ( ρ F / ρ o )1 / 2 ] 2 (5.5) Using the dimensionless groups defined above, two regimes can be identified: a buoyancydominated regime that is correlated by the expression: 69 30 3 L* V* 20 L* V* 2 10 1 0 0 1 4 7 Frf 10 Figure 5.13 HiTAC Flame length and volume correlated with the flame Froude number L* = 8.22 Fr f0.4 (1 + 0.07 Fr f2 ) 0.2 Fr f < 3 (5.6) and a momentum-dominated regime by the expression: L*=11 Fr f ≥ 3 (5.7) Comparing with Delichatsios’ correlation [73] obtained for conventional combustion, L* = 13.5Fr f0.4 (1 + 0.07 Fr f2 ) 0.2 L*=23 Fr f ≥ 5 Fr f < 5 (5.8) (5.9) 70 The criteria constants, L* and Fr f , in the obtained correlation Equ.5.6 and Equ.5.7 are smaller comparing that in Equ.5.8 and Equ.5.9, respectively. This can be understood from their definitions by Equation 5.4 and 5.5, respectively. When the temperature of the oxidizer is increased and the oxygen concentration is decreased, both fs and ρo are decreased. Therefore, L* could be greatly decreased. The criteria constants vary from 23 to 11 at momentum-dominated regime, and the constant at buoyancy-dominated regime varies from 13.5 to 8.22. Influences of oxygen concentration and temperature in oxidizer on Fr f can also be directly analysed as well if ΔT / T∞ in Equation 5.2 is substituted by Equation 5.1. Generally, the decrease of oxygen concentration leads to decrease of Fr f . However, the increase of oxidizer temperature causes the increases in Fr f . The result of both of these compensative effects leads to a small decrease of Fr f . Thus, the criteria Fr number separating buoyancy or momentum-dominated regime varies from 5 to 3. More should be said that there are other similar correlations ( for example in work [75] ) to that of Delichatsios have been proposed for normal combustion conditions. In these works, various investigators use different definitions and experimental techniques to determine flame lengths. However, the effects of oxygen concentration and temperature on these correlations should be similar as the above work. The predicted data of flame volume collapsed to a single linear curve for buoyancydominated regime as: V * = 0.17 Fr f + 0.15 Fr f < 3 (5.10) And a momentum-dominated regime where the dimensionless flame volume V* is constant: V*=0.7 Fr f ≥ 3 (5.11) 71 5.8 Conclusions Influences of combustion with high temperature and low oxygen concentration oxidizer on flame length and volume were numerically investigated. The following conclusions can be drawn: Flame length increases with either the decrease of oxygen content, or increase of oxidizer temperature, or decrease of fuel temperature. Furthermore, the flame length is independent of the fuel flow rate and the diameter of the fuel nozzle. Flame volume increases with either the decrease of oxygen content and increase of oxidizer temperature, or the reduction of fuel temperature, or the increasing in fuel firing rate. Flame volume depends very much on the oxygen concentration in the oxidizer. Influences of high temperature and low oxygen concentration in the oxidizer on the flame Froude number, Frf were examined. Regimes of momentum- or buoyancy-control, were determined on the assumption that oxidizer temperature and oxygen concentration are changeable. A simple correlation of the ‘flame’ length and volume with flow parameters has been derived in terms of a flame Froude number for momentum-buoyancy transition jet flame under the HiTAC condition. The criteria constants of the dimensionless flame volume V* and the dimensionless flame length L* to assess momentum– or buoyancy–control flame are given. 72 6 Flame Entrainments study in coflow single jet flame The entrainment of the jets is the key technique solution for the industrial applications of this novel combustion technology. However, little knowledge is available about the influences of preheated temperature and oxygen concentration on the jet-flame entrainment rate. When combustion occurs in the condition of high-temperature and oxygen deficient oxidizer, reaction zones are relatively widely distributed to yield a somewhat widespread and mild heat release, and hence a uniform temperature distribution. This also implies that the buoyancy force is decreased. The effects of the mild heat release and less buoyancy force on the entrainment could be of fundamental interest. In the previous work, a ‘chemical’ flame is used to describe this less luminosity (or invisible) chemical reaction zone, which is coincident with the internal zone in the outerside border flame. The flame’s outerside border is determined according to the flammability limits of the combustible gases in the combustion chamber in term terms of the oxidation mixture ratio. The length and volume of the ‘chemical’ flame zone are used to describe the physical properties of the chemical reaction zone in the combustion chamber, which have been systemic studied in the previous work. In the context of this paper, the flame means a chemical flame zone. In this work, studies were performed to understand the entrainment into a ‘chemical’ flame zone induced by a turbulent jet flame using the high-temperature and oxygen deficient oxidizer. The combustion with High-Temperature (above the fuel’s auto ignition temperature) and Oxygen Deficient atmosphere is called as HiTAC combustion hereinafter, since those two parameters are unique characteristic differing from any other combustion. A flame entrainment rate is proposed to describe the entrainment induced by a jet flame. The effect of the preheat temperature and the oxygen concentration in the oxidizer, heat release and buoyancy on the entrainment rate is investigated. A correction Richardson coordinate, where the effect of the oxygen concentration (stoichiometric ratio) is included, was derived to describe the local influence of buoyancy force along the chemical flame length under the HiTAC condition. The global behaviour of the entrainment was revealed. Corrections of entrainment rates were derived in terms of an Frf number for momentumbuoyancy transition jet flame under the HiTAC condition. 73 6.1 Study Cases Variables chosen for numerical and theoretical studies were as follows: flow rate and temperature of fuel, oxygen concentration and temperature of the oxidizer, and diameter of the fuel nozzle (Table 6.1). The compositions of the oxidizer (O2, CO2, H2O and N2) were obtained according to the level of oxygen based on the chemical balance of the conventional burner that was used to produce the oxidizer. Table 6.1 Values of Variables of Numerical Studies Cas [O2], To, Tf, Cp1, uF,, u o, es mass% K K kJ/(kg.K) m/s m/s dF, m Reo Lf2, m 105Ris Frf, (wet) 1 5 1173 299 1.3030 25 0.98 5.0E-4 3316 0.625 1.89 0.67 2 10.2 1173 299 1.2653 25 0.98 5.0E-4 3316 0.400 1.90 1.37 3 13.4 1173 299 1.2446 25 0.98 5.0E-4 3316 0.350 1.90 1.83 4 16.9 1173 299 1.2215 25 0.98 5.0E-4 3316 0.313 1.91 2.29 5 23.2 1173 299 1.1790 25 0.98 5.0E-4 3316 0.263 1.92 3.16 6 13.4 1073 299 1.2237 25 0.98 5.0E-4 3316 0.338 1.82 1.73 7 13.4 1373 299 1.2373 25 0.98 5.0E-4 3316 0.375 2.06 1.99 8 13.4 1473 299 1.2228 25 0.98 5.0E-4 3316 0.375 2.13 2.09 9 13.4 1573 299 1.1919 25 0.98 5.0E-4 3316 0.388 2.20 2.18 10 10.2 1173 710 1.2653 25 0.98 7.7E-4 2146 0.425 1.89 1.36 11 10.2 1173 1121 1.2653 25 0.98 9.7E-4 1708 0.438 1.89 1.33 12 16.9 1173 299 1.2215 50 0.98 5.0E-4 6631 0.325 0.48 4.35 13 16.9 1173 299 1.2215 100 0.98 5.0E-4 13263 0.313 0.12 8.49 1 2 Cp is calculated according to the compositions of the oxidizer. Lf was obtained from previous work. Studied temperatures of the oxidizer are greater than 1000K, which is higher than the autoignition temperature of most of fossil fuels, and it is in the range of HiTAC conditions. The coflow speed is about 4% of the jet speeded for most of studied cases. Additionally, since for all studied cases, this velocity was kept as constant, thus its influence was ignored. In addition, the axial distance, x, is counted from the virtual origin from the nozzle. 74 There is no need for a pilot flame to stabilize the flame under HiTAC conditions. This is very important for the studies of the effect of the heat release on the entrainment. Although the mass flow rate of the pilot is very small, the heat release from the pilot occupies a large portion if, for example, the hydrogen is used as the pilot fuel. The heat-flux ratio between the jet and pilot is only 22% even the pilot-to-jet mass flux ration is less than 1.2%, as analysed in the work [90]. Therefore, in the HiTAC condition, the studies of the effect of the heat release on the entrainment especially in the near field will have a higher accuracy. Flame length was calculated as the distance between the end of the fuel nozzle and the axial location of the oxidation mixture ratio equal to 0.99. The lift-off distance was negligible. There will be no any error for the calculation of flame length. In this work, a flame entrainment rate, Rent , is proposed. It is defined as a ratio of the mass flow rates through the cross section of the flame (me) to the initial jet mass rates (m0), and be expressed as: Rent = me m0 (6.1) Here, d * = d F ( ρ F / ρ o ) 0.5 (6.2) The mass flux, me, can be calculated by the following integral of velocity and density profiles for whole section of the chemical reaction zone. me = ∫ R flame 0 2 ρu πrdr = ∑ ρ i u i Ai (6.3) i is the calculation cell number. This definition of this flame entrainment rate can be used to estimate the flame entrainments by a group of jet flames, which can be interference between the individual jet flames. In this paper, only one jet flame is studied. Consequently, the results from this work can be compared with that from the previous published works, which only single jet 75 entrainment was studied. In order to compare with the available data from the literatures, it is deducted as the well-know expression: Rent = me m0 = C e ( ρ o ρ F ) 0.5 x d F = C e x / d * (6.4) 6.2 Effects of the oxygen concentration of the oxidizer on entrainment The entrainment rate for the burning region of jet flames with different oxygen concentration in the oxidizer are presented in Figure 6.1. 120 Case1 Case2 Case3 Case4 Case 5 me/m0 80 40 Chemical Flame length 0 0 100 200 300 400 500 600 x/d* Figure 6.1 The variation of the entrainment rate along the axial direction from fuel nozzle tip at different oxygen concentrations in the oxidizer It can be seen that the curves show that the trend is similar at different oxygen levels. The entrainment increases linearly for the first half of the flame length, reaching a maximum value, and the decrease of the entrainment occurs over the last third of the flame length. 76 This global various trend might be explained by the nature of the jet. From previous studies [88], jets in co-flow are known to show a jet-like behaviour in the near field, but a wakelike behaviour in the far field. For a fully developed self-similar reacting jet, the entrainment coefficient is a constant. For a fully developed wake with a small excess velocity, the entrainment is achieved by the encroachment of the boundary on the surrounding fluid, and should decrease according to self-similarity. This can be observed in the jet-like region (from the nozzle tip to the location of the maximum value of the entrainment rate) in Figure 6.1, where the entrainment rate increases, and the wake-like region (from the location of the maximum value of the entrainment rate to the end of the chemical flame), where the entrainment decreases, and the entrainment coefficient has a negative value. There are few reports in the literature about this phenomenon for the reacting jet. One of the most important reasons is the used definition of the flame border. In this work, the chemical flame border is separated even at the tip of the flame. At far downstream locations in the reacting jet, the distribution becomes smaller than the free jet. Therefore, a reduction of the entrainment is possible. It is worthy to note that both the near and far fields of the jet are of importance. In the near field, a significant portion of the mixing process occurs, which affects the flame stabilization, and in the far field, the flame length is defined. In particularly, the location of the maximum entrainment decides the optimal design of the furnace and the location of the burners. In this part, these behaviours of the entrainments are discussed firstly, and the location of the maximum entrainment will be discussed later. The average entrainment coefficients Ce from the nozzle tip to the locations before the maximum entrainment for Case 1, 2, 3, 4, and 5 were summarized in the Table 6.2. The entrainment coefficient Ce was determined to be 0.32 for momentum driven free jet by Ricou and Spalding [76]. Becker and Yamazaki [78] reported Ce =0.16 for a momentum driven reacting jet. Han and Mungal reported [82] Ce=0.13 for a reacting free jet. A summary of the results from the present study and previous reported work are listed in Table 6.2. 77 Table 6.2 Summary of the results for present study and previous works Reo [O2], Flame uF, m/s uo,m/s Lf/dF Ce mass% Case 1 3316 5 Yes 25 0.98 1250 0.28 Case 2 3316 10.2 Yes 25 0.98 800 0.20 Case 3 3316 13.4 Yes 25 0.98 700 0.17 Case 4 3316 16.9 Yes 25 0.98 625 0.15 Case 5 3316 23.2 Yes 25 0.98 525 0.14 R&S [76] >25 000 23.2 No * 0 * 0.32 Muniz &Mungal 10 000 23.2 Yes 36 0.45 100 0.12a &Mungal 37 500 23.2 Yes 134 2.0 152 0.076a 23.2 Yes 0.16b 23.2 Yes 0.13(1-e-0.036(r-1)) [85] Muniz [85] Becker & Yamazaki [78] Han &Mungal [90] a Effect of ρi′ and ui′ is neglected. Ce could be underestimated by around 20%. b For the entrainment coefficient at small ξ. One major difference in the present work and other reported results is the conditions of the oxidizer. Here, high temperature and oxygen deficient oxidizer was used. Another difference in the present and other results is the definition of the entrainment, which the flame entrainment is used in this work rather than the jet entrainment in other works. It can be observed that the entrainment rate for the studied cases are in the range of 0.14 ≤ C e ≤ 0.28 . They are smaller than those of non reacting jets, but larger than those of reacting jets under normal combustion conditions. It is been known from previous studies [78, 82, 85], that heat release in reacting jets reduces the entrainment. This means that the entrainment coefficient for the reacting jets should be lower than 0.32. This principle is also valid by flames under the HiTAC condition. Table 6.2 shows that the maximum entrainment rate for the studied cases occurs at the lowest oxygen concentrations. With the reduction of the oxygen concentration, the 78 entrainment increases. For example, the flame at the case of 5% oxygen concentration experiences a 50% increase in the entrainment than at 21% oxygen concentration. This trend is consistence with the variation of the HiTAC flame volume or temperature uniformity. It has been shown, that a larger chemical reaction zone exits when the combustion occurs in a high temperature and low oxygen concentration oxidizer. Lower the oxygen concentration in the oxidizer, larger the chemical reaction zone and consequently a more uniform the heat release. Therefore, these features might be used to explain the effect of the oxygen concentration on the entrainment. i.e., a larger heat release uniformity, results in a larger entrainment. In order to understand the present results, further analysis of the effect of heat release on the entrainment rate is necessary to understand the above results. In the absence of buoyancy, the momentum flux of a turbulent jet, which is defined as the momentum crossing a plane perpendicular to the jet axis, remains constant with axial distance and is equal to the initial jet momentum flux. When ambient fluid is entrained into a turbulent jet, the mass flux of the jet increases and the velocity of the jet decreases due to the conservation of the momentum. It has been shown [87], that flames have higher centreline velocities and concentrations than those of non reacting jets at the same axial location with the same exit conditions. It can be assumed that the total mass flux is smaller in a flame than in a non reacting jet with the same exit conditions. Therefore, the mass flux by entrainment is less if the initial mass fluxes are the same. The increase in centreline velocities and concentrations is caused by the heat release. It can be concluded that a uniform heat release rate will have a less effect on these parameters, thus a larger entrainment rate. It has been demonstrated [2,14-20] that the concentration and temperature profiles are more uniform under the condition of high-temperature and oxygen deficient oxidizer than those in the conventional condition. Furthermore, a recent study proved [51] that the centreline velocities are smaller when the oxygen concentration in the oxidizer decreases during the HiTAC condition. These features implicates that the entrainment during the HiTAC condition will be larger than under conventional conditions. At the extreme, when the oxygen concentration in the oxidizer is equal to 0, the most uniform heat release is achieved, and there will be no effect by the heat release. Additionally, lower centreline velocities at the lower oxygen concentration lead to a larger the flame widths, which have proved in [2, 14-20]. Consistent with this, the entrainment at the low oxygen concentration increases. 79 The fluctuations of the velocities and temperatures also affect the entrainment rate. These intensities of the fluctuations vary along the flame length. It is known that there exists some effect that increases along the flame length, that enhances the large-scale structures and fluctuations in a reacting jet [82]. Becker and Yamazaki [84] identified this effect as buoyancy, which adds additional momentum to the jet. This will be further discussed later. Finally, one possible reason for the increase in the entrainment rate in HiTAC combustion than that in the normal cases is the use of the different definition of the entrainment. Here, the flame entrainment is used rather than the jet entrainment, which was adopted by most of researchers. In HiTAC combustion, the flame width increases with the reduction of the oxygen concentration [2, 14-18].This contributes the increase of the entrainment as well. One effect the reduced entrainment has on the flame is an increase of the chemical flame volume. Consequently, an estimate of the entrainment increase caused by lower oxygen concentration in the oxidiser can be obtained from the flame volume growth rate, because the flame volume is dependent on the oxygen concentration [2,14-18]. Here, the effect of oxygen concentration on the entrainment rate has been shown in Figure 6.2a. It can be seen that the entrainment increases when the oxygen concentration in the oxidizer reduces. This is clearer if the oxygen concentration in the oxidizer is below 10%. This trend is consistent with changes of the chemical flame volume versus the oxygen concentration in the oxidizer. Moreover, this trend becomes weaker when the oxygen concentration is approach to 21% in the oxidizer. 80 Entrainment coefficient, Ce 0,4 0,3 0,2 0,1 0 0 5 10 15 20 25 Oxygen Concentration in the oxidizer, % Entrainment coefficient, Ce a) 0,4 0,3 0,2 0,1 y = -0,6306x + 1,0099 0 1 1,1 1,2 1,3 1,4 1,5 (Tf/To)**0.5 b) Figure 6.2 Entrainment rate change versus the oxygen concentration and the characteristic temperature ratio (T f / To ) 0.5 . 81 The uniformity of the heat release can be simply represented by the characteristic temperature ratio, (Tf/To). Then, the effect of uniformity of the heat release on the entrainment can be further quantitative analyzed. It is very obvious that the highly preheating temperature and oxygen concentration of the oxidizer plays an important role on the flame temperature during the HiTAC combustion. Consequently, the HiTAC’s flame temperature can be featured when it is expressed as a function of the preheated temperature of the oxidizer. The characteristic temperature ratio can be estimated through the relationship from previous work as following. T f ,max = To + Tad 21 +1 [O2 ] (6.5) and use Tf,max instead Tf, then, the characteristic temperature ratio can be obtained: Tf To = Tad 1 +1 To 21 +1 [O2 ] (6.6) Here Tad is the adiabatic temperature of propane at the stoichiometric combustion with air (79% N2 and 21% O2). The entrainment rate in this work is plotted as a function of the characteristic temperature ratio resulting from combustion, (T f / To ) 0.5 in Figure 6.2 (b). It can be seen that the entrainment rate changes with the same factor of the (T f / To ) 0.5 when (T f / To ) 0.5 is less than 1.4. It is interesting to note that Han [82] argued that in the near field of reacting jets, the effect of heat release on the entrainment can be estimated by the ratio of (ρo/ρa)0.5. The ratio of (ρo/ρa)0.5 can be approximated by (Tf/To)0.5 for a constant pressure condition. This result is consistent with our presented data. 82 To summarize of these discussions, the conclusion can be made gives on the hand that the entrainment rate during high-temperature and oxygen deficient conditions will be larger than under conventional combustion. The lower the oxygen concentration in the oxidizer is, more uniform the heat release rate is, thus larger the entrainment rate. The effect of heat release reduces the entrainment in the near field of the reacting jets with the same factor of the characteristic ratio (Tf/To)0.5. 6.3 Effects of the temperature of the oxidizer on entrainment The preheated temperature of the oxidizer was increased when the other conditions were kept constant for the studied cases (case 3, 6-9). Preheat temperature of the oxidizer has small effect on the entrainment for studied cases, as shown in Figure 31 60 Case 3 Case 7 Case 9 Case 6 Case 8 50 100 me/m0 40 20 0 0 150 200 250 300 x/d* Figure 6.3 The variation of the entrainment rate along the axial direction from fuel nozzle tip at different preheated temperature of the oxidizer 83 The studied cases have the same entrainment ratio up to x/d*<50. This trend can be extended to the point of the maximum entrainment ratio. At the flame tip, a slightly higher entrainment takes place at lower preheat temperatures. The momentum of the fuel jet for these cases was maintained constant. As the inlet temperature of the oxidizer is increased, the mass flux by turbulent mixing is reduced due to the decrease of air density. Meanwhile, the small expansion of the burnt gases does not suppress the turbulent mixing. These effects are considered to compensate each other. As a result, the entrainment is maintained almost constant. In particular, up to x/d*<50, it can be seen that this is the jet developing regime. The same momentum will cause the same entrainment. At the flame tip region, located in the buoyancy –control zone, the coherent structures of the flame are easy to penetrate the ambient gases. This result can also quantitatively explained by using the above relationship: Re ∝ (T f / To ) 0.5 . When the preheat temperature increases for the same level oxygen concentration, the flame temperature increases as well. Thus, the characteristic temperature ratio has a small variation, and this variation becomes smaller when the oxidizer’s temperature is increased. Consequently, the effect of the preheating temperature of the oxidizer is smaller, especial in the momentum controlled zone (near field). In the far field, it might be in the regime of the buoyant force- control zone. 6.4 Effects of the fuel flux on entrainment For the studied cases 4, 12 and 13, the fuel fluxes were increased through an increase of the fuel velocities, while the other conditions were kept constant. The effects of the fuel flux on the entrainment were shown in Figure 6.4. The increase in the fuel fluxes leads to a slight increase in the flame temperature, the density of the flue gases is decreased, thus the low entrainment. Additionally, the diameter of the flame obviously increases as the fuel flow rate increases which increases the mass entrainment. These factors compensate each other and achieve a constant mass entrainment. 84 60 Case4 Case 12 Case 13 me/m0 40 20 0 0 50 100 150 200 250 300 x/d* Figure 6.4 The variation of the entrainment rate along the axial direction from fuel nozzle tip at different preheated temperature of the oxidizer 6.5. Effects of the buoyancy Becker and Yamazaki [78] have established a criterion by which the effects of buoyancy in a flame can be quantified. The parameter of interest is a nondimensional stream wise coordinate, ξ ξ ≡ Ris1 / 3 x ds (6.7) where, Ris is the source Richardson number, and ds is source diameter. For a uniform exit velocity, uo, the source Richardson number and diameter become: Ris = gds/uo2, where g is acceleration due to gravity. ξ contains the integrated effect of the buoyancy along the jet trajectory. Buoyancy forces are considered to be negligible when ξ< 1. This coordinate is widely used by other researchers. [82, 85] etc. 85 It should be noticed that this relationship has not considered the influences of the oxygen concentration and the air preheated temperature. This is because of an assumption, that ρf is typically less than 0.5ρo was used when this relationship was derived. This assumption is reasonable for the combustion conditions without preheated air. However, it will cause a large error when this assumption is used in combustion under highly preheated and oxygen deficient conditions in the oxidizer. Therefore, at least a correction of this coordinate should be added to consider the influences of the preheated temperature and oxygen concentration. A simply approach is introduced as follows. If we consider a cone-shaped jet with the width, δ, the ratio of the buoyancy that the entire jet brush feels (Fb)compared to the initial momentum (Fm)can be written as [82], Fb = Fm (ρ o − ρ f )g ρFu 2 F π 4 π 4 d δ2 x 3 (6.8) 2 F Letting δ ∝ x and applying a cubic root to the entire expression, the nondimensional stream wise coordinate, ξ ' can be wrote as: ξ ' ≡ Ri01 / 3 ρf x (1 − ) * ρo d (6.9) here, Ri0 = gd * / u F2 (6.10) If we assume ρf is typically less than 0.5ρo, i.e., ( ρ o − ρ f ) → ρ o , we can obtain the expression as Equation 6.9, which defined by Becker and Yamazaki [78]. Under HiTAC conditions, this assumption is not reasonable. Further analysis is required. 86 In order to obtain a more concise expression of the nondimensional stream wise coordinate, ξ’, the relationship between ρf and ρo has to be addressed. Considering the relationship between gas densities and its temperatures, one has: ρf T ∝ o ρo T f (6.11) Combing Equation 6.11 and Equation 6.6, one can obtain: ρf ∝ ρ o Tad 1 1 +1 To 21 +1 [O2 ] (6.12) Thus, Equation 6.12 can be modified as: ξ HiTOD ≡ c HiTOD Ri01 / 3 x d* (6.13) Here, CHiTAC can be named as the Richardson correction ratio for the HiTAC combustion, and c HiTOD ≡ 1 To 21 1+ ( + 1) Tad [O2 ] (6.14) From Equation 6.14 it is easy to find the effects of the assumption, ( ρ o − ρ f ) → ρ o even for the normally combustion (21% [O2]). In this case, Equation 6.414 can be conversed as: c ≡ 1 2T 1+ o Tad (6.15) 87 If the adiabatic flame temperature is significantly larger than the free stream (ambient) temperature, i.e. c=1, the effects of the stream temperature is neglectable. Otherwise, this assumption could not be reasonable. The effect of oxygen concentration and preheated temperature of the oxidizer is considered in the corrected Richardson number. Here, buoyancy forces are still considered to be negligible when ξHiTAC< 1. In Figure 6.5a, ξHiTAC in log scale is plotted as a function of axial distance from the jet origin for the different caes. In order to understand the effect of the buoyancy forces on the flame length, ξHiTAC is presented in log scale as a function of the axial distance from the nozzle, normalized by the flame length, Lf in Figure 6.5b. Figure 6.5 shows the influence of buoyancy on the HiTAC flame. It can be seen that the buoyancy force increases with the reduction of the oxygen concentration whilst comparing with Case 1, 2, 3, 4 and 5. For Case 5 (21%O2), the buoyancy forces are considered negligible for z/d*< 72. The value for case 1 (5%O2) is 140, which is approximately twice the length over which Case 5 is non-buoyant. This can be understood as a more uniform temperature profile provides smaller buoyancy forces. Furthermore, for Case 5, this criteria location is 0.4 of the flame length, which means from the nozzle tip up to 40% of the flame length; it is in the momentum-control regime. This value is 30% for Case 1. This implies that more of the flame’s length locates in the momentum –control regime for higher oxygen concentration as well. 88 100 Corrected Kesi Buoyancy non-negligible 1 Case1 Case 3 Case 5 Case 9 Case 11 Case 13 Momentum-driven Case2 Case4 Case 7 Case 10 Case 12 0.01 0 100 200 300 400 500 600 x/d* (a) Corrected Kesi 10 1 0.1 Case1 Case2 Case 3 Case4 Case 5 Case 7 Case 9 Case 10 Case 11 Case 12 Case 13 0.01 0.0 0.2 0.4 0.6 0.8 1.0 x/Lflame (b) Figure 6.5 The local influence of buoyancy as a function of the axial distance normalize by a) (x/d*), and b) the flame height in log scale for difference cases 89 The influence of the preheat temperature of the oxidizer on buoyancy is minor for the studied cases (Case 3, 7, and 9). The buoyancy force can be considered negligible for 0.35 of the flame length for these cases. This also implies that the effect of the oxidizer temperature on the temperature field uniformity is small when the oxidizer’s temperature is higher than 1000K. This is because of the increase of the air preheat temperature also causes an increase in the flame temperature, the different flue gas densities, which is the major factor for affecting the buoyancy force, is almost kept constant. The fuel fluxes have stronger effects on the entrainment. The momentum driven regime has an increase in the length with the increase of the fuel fluxes. For example, the effects of buoyancy become non-negligible for x/d* >82 for Case 4. These values for case 12 and 13 are 125 and 200, respectively. It is obvious that the larger the initial monument is, it maintains a longer momentum control regime, and the Ris decreases as the factor of (uo)2. It can also be seen from Figure 6 that the larger initial monument causes a longer flame length expose in the monument control regime. For the case 13, around 80% flame length locates in the regime of monument - controlling. The values for case 12 and 4 are 45% and 32% respectively. 6.6. Global field behaviour of the entrainment We have detail discussed the entrainment behaviour in near field. In this part, the global field behaviour of the entrainment is further chased. It was chosen to present the data as a function of the axial distance(x) normalized by the flame height, Lf, because the general relationships of the flame length have been well studied. Here, Lf is the chemical flame length can be obtained in previous work. Another dimensionless parameter, m* is adopted. It is an expression of an entrainment rate normalized by the stoichiometric requirements: m* = me ( S + 1)m F (6.16) here, 90 S = (mo / m F ) stoic (6.17) Figure 7 shows the relationship of this dimensionless parameters, m*, and the parameter (x/Lflame). In order to see clear various, the log scale of the distance axial is used. It can be seen that all the maximum normalized entrainment positions can be collapsed to a single point. It appears at the location of x / L f ≅ 0.7 . Basing on this point, two regimes can be identified as a near and a far field. m*=me/((S+1)mF) 2 1,5 Case 1 Case 5 Case 9 Case 13 Case 2 Case 6 Case 10 Case 3 Case 7 Case 11 Case 4 Case 8 Case 12 1 0,5 0 0,001 0,01 0,1 1 10 x/Lflame Figure 6.6 Dimensionless entrainment rates v.s. ratio of axial distance from the nozzle tip to flame height From Figure 6.6, the mass entrainment of the near field can be expressed as: me x = 1.77 Lf ( S + 1)m F x / L f ≤ 0.7 (6.18) and the mass entrainment of the far field can be stated as: 91 me x x = 3.67 − 3.31 = 3.31(1.11 − ) Lf Lf ( S + 1)m F x / L f > 0.7 (6.19) Obviously, the entrainment coefficient is positive in the near field where and it is negative in the far field. It is possible to use available flame length’s relationship to further simply above relationships. In this work, the relationship of the flame length by mean of the flame Froude number during the HiTAC condition obtained in previous work is used. Only the entrainment in near field, x / L f < 0.7 , is discussed in this part. According to the results from previous work, the flame length during the HiTAC condition can be decided by the flame Froude number as: L* = 11 L* = Fr f ≥ 3 8.22 Fr f0.4 Fr f < 3 (1 + 0.07 Fr f2 ) 0.2 (6.20) (6.21) Here, L* = [d Lf fs F (ρ F / ρo ) 1/ 2 ] = Lf * (6.22) 1/ 2 (6.23) d ( S + 1) and Fr f = u F f s3 / 2 ⎡ ΔT f ⎤ ρ ( F )1 / 4 ⎢ gd F ⎥ ρ∞ ⎣ T∞ ⎦ fs = 1 (mo / m F ) stoic + 1 (6.24) 92 Then, a simplification of the entrainment in the near field can be obtained through substituting Equation 6.20-Equation 6.24 into Equation 6.18. They can be expressed as: me x = 0.16 * for Fr f ≥ 3 mF d 2 0.2 me (1 + 0.07 Fr f ) x = 0. 4 mF 4.64 Fr f d* for Fr f < 3 (6.25) (6.26) It can be seen from Equation 6.25, the entrainment coefficient, Ce is equal to 0.16 for a momentum driven reacting jet ( Fr f ≥ 3 ), which is comparable with the data by Becker and Yamazaki [78]. These relationships can also be use to compare the entrainment coefficient which were obtained in the previous part in this paper. For example, Case 12 and 13 belong to the momentum-dominated regime (see Table 1) because of their Frf numbers are greater than 3. Then, the entrainment coefficient, Ce, of these two cases is 0.16 according to the Equation 6.25. This value is only 6% larger than that obtained in previous data (Ce=0.15). A good agreement can be obtained. It is also very interesting to calculate the maximum mass entrainment, which can decides the optimal design of the furnace and the location of the burners. Let’s put x / L f = 0.7 into Equation 6.25, then the maximum entrainment can be estimated as: me = 1.24( S + 1) mF (6.27) Comparison the maximum entrainments calculated by Equation 6.27 and the results presented previous part, the results are also very encouraged. 93 6.7 Conclusions The influences of combustion using a high-temperature and oxygen deficient oxidizer on the flame entrainment induced by a turbulent reacting jet are numerically and theoretically investigated. A flame entrainment ratio is proposed. The results can be summarized as follows. 1. The uniformity of the heat release in reacting jets has strong effect on the flame entrainment. More the uniformity of the heat release, larger the entrainment. The effect of heat release reduces the entrainment in the near field of the reacting jets with the same factor of the characteristic ratio (Tf/To)0.5. 2. The entrainment increases as the oxygen concentration is decreased. Furthermore, the entrainment is independent of the fuel flow rate and the preheated temperature of the oxidizer for the investigated temperature range (1073-1573K). 3. The effect of the oxygen concentration and preheated temperature of the oxidizer on buoyancy was examined. A correction Richardson coordinate, ξ HiTAC ≡ c HiTAC Ri01 / 3 x , d* where the effect of the oxygen concentration (stoichiometric ratio) is included, was derived to describe the local influence of buoyancy along the axial distance from the nozzle under the HiTAC condition. It can be concluded that the buoyancy force increases with the reduction of the oxygen concentration in the oxidizer. 4. The global behaviour of the entrainment was revealed. Two regimes for the entrainment have been identified in jet flames (a) the near field where entrainment rates are positive; and (b) the far field where entrainment rates are negative. Corrections of entrainment rates were derived in terms of a Frf number for momentum-buoyancy transition jet flame under the HiTAC condition. Furthermore, the maximum entrainments along the flame length are estimated. 94 7 Semi-industrial furnace with HiTAC burners study 7.1 Experimental measured and verification of mathematical modelling for HiTAC furnace with one-flame burner Experimental measurement and modelling validation are performed by the means of the following data: - Energy balance, - Wall temperature profiles, - In-furnace gas species These are done using tests based on the HiTAC test furnace equipped with one-flame Highcycle Regenerative System (HRS). 7.1.1 Energy Balance The reference temperature is set at T=298 K, so the sensible heat flow rate at the fuel inlets was zero. The overall thermal energy (only including the fuels’ chemical energy) input to the test furnace in this study was 182 kW. The thermal input of the fuel for the simulation was calculated according to the chemical reaction steps (R3.1-R3.5). The heat of exhaust after the burner for simulation was approximately calculated as the value of heat of flue gas through the burner minus combustion sensible heat of the combustion air. The summary of the energy balance of the semi-industrial furnace at the operation condition is given in Table 7.1. A figure of 54.65% of the predicted fuel thermal input passed through the burner, and 83.6 percent of the sensible heat carried by the flue gases through the burner outlets was used to preheat the combustion air from 300 to 1211 K. This means the thermal efficiency of the regenerator is 83.6%. This value is about 45.71 % of the total thermal input. This implies that a very high level of energy utilization efficiency is achieved. Reiterating here, 8.94% 95 of the predicted total fuel thermal input is removed by flue gas through the burner. This value is higher than measured value of 5.21%. Possible reasons for this could be one or a combination of the following: heat loss in the burner, or the measurement point being on the outside of the burner. The predicted amount of heat taken by the air-cooling tubes occupies 51.67 % of the fuel thermal input. The heat absorbed by air-cooling tube was measured to be around 54.27% of the fuel thermal input within the margin of measurement error of 6.44%. Therefore, the predicted and measured amounts are in a reasonable agreement. The predicted results also indicated that 97.4% of the heat transferred to the air-cooling tube was due to radiation, and 2.6% was due to convection. Furthermore, the calculation enthalpy of the chimney flue gas constitutes 9.78% of the fuel thermal input, while the actual value measured was 11.25%. Again, the agreement is acceptable. The predicted heat loss through the furnace walls accounts for 29.61% of the thermal input, compared with an actual measurement of 29.49%. Thus, these are also in a good agreement. Table 7.1 Fuel characteristic and burner operating conditions HiTAC mode Fuel flow rate [Nm3/h] 7.7 Fuel inlet temperature, [K] 298 Fuel composition (Mass fraction, [%]) CH4 (0.02), C2H6 (0.95), C3H8 (98.35) , C4H10 (0.67). Combustion air flow rate, [Nm3/h] 200 Combustion air temperature,[K] 1211 K Combustion air composition ([Vol %]) N2 (79), O2 (21) 96 Table 7.2 Energy balance from measurements and from modelling data (Reference temperature: 298K) Energy [kW] 1. Fuel power Input 2. Combustion air sensible heat Total 3 Heat taken away by the air-cooled tubes HiTAC trial measured 184.04 182.0 84.12 (after preheated by the 1.85 (before preheated by burner) burner) 268.16 183.85 95.09 99.78 4 Heat of flue gas through burner 100.58 5 Heat of exhaust after burner 16.46 9.57 17.99 20.69 51.27 48.13 3.23 -- Uncounted loss -- 6.06 Total 267.85 183.85 6 Sensible heat of flue gas through Output HiTAC CFD prediction main chimney 7 Heat loss from walls Radiation heat loss at inlets and outlets It is possible to improve the efficiency of the furnace by further increasing the efficiency of the heat recovery from waste flue gases from the current value of 80% to 100%. The extra heat recovery from the flue gases can be used to preheat the fuel, which can bring benefits, such as an even greater reduction in NO emissions. This is very important for combustion stability when using low and medium caloric value fuels in HiTAC technology as in these cases, the fuel volume is larger. This method features preheating of both fuel and air, and is referred to in the field as twin-preheating. As such the method involving preheating of the combustion air only can be referred to as single-preheating. 7.1.2 Temperature Field 97 Furnace temperatures in the HiTAC test furnace were measured at various positions along the left-hand-side wall of the test furnace (viewed from the burner). Figure 7.1 presents a comparison of temperature predictions and measurements showing reasonable agreement, with a maximum difference of about 10 K with a range of error of 2.8%. These values were obtained from a stationary thermocouple located on the furnace wall. 1500 Temperature, oC 1400 Measured Prediction 1300 1200 1100 1000 900 800 0 400 800 1200 1600 Z (mm) 2000 2400 2800 Figure 7.1 Temperature distribution on the side wall of the furnace at x = 0.8 m, and y = 0.3m 7.1.3 Gas Species Figures 7.2 and 7.3 present the measured and predicted concentrations of O2 and CO across the furnace chamber at specific vertical distances from the burner face. The x-axis represents the vertical distance from the centreline of the burner. From figure 7.2, it can be seen that the calculated O2 levels are in good agreement with that measured. Both the measured and predicted curves show the same locations and magnitudes for the maximum and minimum O2 and CO2 concentrations. Furthermore, the measured CO concentration agrees with that predicted. The predicted CO concentrations on, or close to, the burner centerline are lower than the measured values, however, the 98 relative difference decreases with increasing distance from the burner. Meanwhile, the modelling underestimates the fuel consumption in the front part of furnace. From these data, the conclusion can be drawn that the flame diffusion in furnace is well predicted since there is good agreement of the locations and magnitudes of the combustibles, including O2(dry%) hydrocarbon, and CO and O2. 12 10 8 6 4 2 0 -800 Measured predicted -400 0 Y (mm) 400 800 (a) 12 O2 (%dry) 10 8 Measured predicted 6 4 2 0 -800 -400 0 Y (mm) 400 800 O2 (%dry) (b) 12 10 8 6 4 2 0 -800 Measured predicted -400 0 Y (mm) 400 800 (c) Figure 7.2 Predicted and measured O2 profiles in the furnace (a) x = 0, z = 0.3 m (b) x = 0, z = 0.6m (c) x = 0, z = 1.2 m 99 CO (ppm) 16000 12000 M. P. 8000 4000 0 -800 -400 0 Y (mm) 400 0 Y (m m) 400 0 Y (mm) 400 0 Y (mm) 400 800 (a) CO (ppm) 25000 20000 15000 M P 10000 5000 0 -800 -400 800 (b) CO (ppm) 16000 12000 M. P. 8000 4000 0 -800 -400 800 (c) 5000 M P CO (ppm) 4000 3000 2000 1000 0 -800 -400 800 (d) Figure 7.3 Predicted and measured CO profiles in the furnace (a) x = 0, z = 0.3 m , (b) x = 0, z = 0.6 m, (c) x = 0, z = 1.2 m, (d) x = 0, z = 2.15 m 100 7.1.4 Features of Combustion and Flow of HiTAC Simulation studies were performed and the differences between the heat transfer and combustion features between a conventional high velocity turbulent jet flame and HiTAC flame were found. The influence of a heat sink on furnace heat transfer with these two types of burner systems was investigated as well. Vectors of the in-furnace gas velocity for HiTAC with one-flame HRS are shown in Figure 7.4. A cross section through the fuel and one of air inlets is presented in order to clearly show the flow characteristic of HiTAC. The combustion air was injected into the furnace with a velocity as high as 100 m/s. This large injection velocity leads to strong internal recirculation zones (IRZ). On the other hand, flue gas flows to the root of flame on its way out through the burner located at the root of flame as shown in figure 7.4. Recirculation allows good mixing of the combustion air with the flue gases before ignition occurs. Because of this, the HiTAC mode leads to a lower peak temperature and a larger combustion volume and consequently a lower NO emission. Thus, the HiTAC flame stability depends on existence of the strong internal recirculation zones. 101 Figure 7.4 Predicted velocity vectors at a cross section through the fuel and one of the air inlets in HiTAC mode [m/s] 102 The highest temperature zone was found along the central axis of the furnace for both analyzed cases (Figure 7.5). The peak temperature zone is further away from the burner face at a distance of about 1.0 m. The maximum gas temperature for the HiTAC mode is lower than that for the turbulent jet flame mode although the combustion air for HiTAC firing mode is preheated up to 1223 K. For example, for the design operation of burner, comparing HiTAC and conventional firing modes, the maximum temperature difference is equal to 361K. This is the result of a very intensive flue gas recirculation created by internal recirculation zones in HiTAC firing mode. Figure 7.5 Predicted temperature profile at a cross section through the fuel and one of the air inlets at HiTAC mode [K] The predicted gas temperature field uniformity is higher for the HiTAC mode, which is a known and expected advantage of the HiTAC technology. However, one should be very 103 careful when using ‘more uniform temperature distribution’ to describe HiTAC performance. This is demonstrated by the fact that the temperature near the burner zone is very low, for example, 870K for one-burner regenerative burner system studied here. This is very much lower than the furnace temperature, thus implying that temperature distribution is not uniform. Figures 7.6(a) shows the predicted flame volume by means of the above definition. Figure7.6(b) gives the flame shape derived from the in-furnace gases measured value according to the same definition. Table 3 shows that both the measured and predicted of the flame geometric and physical characteristics. The flame zone is almost limited to the volume of the hypothetical cylinder created by the air jets because of their strong injection momentum. Consequently it leads to a rather long flame. The furnace flame occupation coefficient was 3.174% for studied case, and this value is 13.1 times for the same fuel capacity with a normal high velocity jet burner. The large flame volume is consistent with the distribution of species. The predicted and measured results are found to agree well with respect to the shape of the flame zone. However the measured flame volumes are slightly larger than those predicted. A possible reason for this is that only half a burner cycle was calculated in order to get the steady-state condition in this regard. In fact, it is quite possible that the mixing of fuel and air is in reality more intense than in the modelling due to the periodical switching of the burner. 104 (a) (b) Figure 7.6 Predicted and measured flame shapes and volumes shown by the oxidation mixture ratio RO=0.99 (a) Predicted (b) Estimated by experimental data 105 The predicted and measured values of the maximum normalized flame lengths are 12.95 and 12.74 respectively, and are calculated as the ratio of flame length to burner diameter as shown in Figure 7.6 (b). The predicted flame length agrees well with the measured length within a 4.3% measured error margin. Moreover, the predicted and measured maximum normalized flame diameters are 2.4 and 2.25 respectively, defined as the ratio of the maximum flame diameter to the burner diameter, as shown in Figure 3.5 (b). This implies that the flame spread is in good agreement. However, while the predicted maximum flame diameter occurs at the end of flame, the measured flame does not exhibit this feature. One of possible reasons for this is that only half a burner cycle was calculated, and another possible reasons for this could be the measurement was not taken at the rear of the furnace (2.6 m from face of the burner), due to the test furnace construction limitation. Figure 7.7 shows the typical difference from a conventional high velocity turbulent jet burner. Comparing with the flame between the flame shapes and sizes for a HiTAC burner (Figure 7.6(a) and a conventional high velocity turbulent jet burner (Figure 7.7). From the figure it is very clear that the HiTAC flame spreads over a much larger volume than the conventional flame. The furnace flame occupation coefficient in the case of HiTAC is 15.8 times larger than for the conventional flame mode. The predicted maximum normalized flame length for the one-flame HRS is 12.95 at the design condition and the predicted maximum normalized flame diameter is 2.4. The heat release zone (chemistry reaction zone) for the HiTAC mode is significantly larger than for the conventional flame mode. By implication, the firing rate in the HiTAC flame is much smaller than that for the conventional flame mode. The maximum firing rate for the flame mode is 1.87×106 kW/m3, which is 48.9 times higher than for the HiTAC mode where the maximum firing rate is 3.82×104 kW/m3. The average flame heat release for the flame mode is also much greater than that for HiTAC mode. This proportion is consistent with the flame volume as shown above. It is a known fact that the flame volume is significantly larger in HiTAC mode for the same type of fuel and the same firing rate. Moreover, the value of firing rate also implies that the combustion noise in HiTAC mode is much lower than in the flame mode since any approach that reduces combustion intensity within a combustion reaction may be expected to reduce the sound power produced by a flame. Further analysis can be found in references made in [118]. 106 (b) Figure 7.7 Predicted flame shape for Conventional burner Figure 7.6b gives the measured flame shape according to the previous definition. The flame zone is almost limited to the volume of the hypothetical cylinder created by the air jets because of their strong injection. Consequently it leads to a rather long flame. The measured flame volume was 0.27 m3, and the furnace flame occupation coefficient was 3.77% for studied case. This value is around 35 times larger than that for the same fuel capacity with a normal high velocity jet burner. The large flame volume is consistence with the distribution of species. The measured value of flame length was 2.8 m, and its normalized flame length was 12.74, which was calculated as the ratio of flame length to burner diameter as shown in Figure 3.5 (b). The flame almost spread the whole furnace length. Moreover, the measured maximum flame diameter was 0.5 m and its normalized flame diameter was 2.25, defined according to the ratio of the maximum flame diameter to the burner diameter, as shown in Figure 3.5 (b). The measured large flame volume also 107 implies that HiTAC has a big chemical reaction zone, and consequently a uniform temperature field. This is a result of very intensive flue gas recirculation created by internal recirculation zones. Therefore, for the same total fuel thermal input in a furnace, the heat release rate becomes more uniform than the normal combustion. Consequently, the maximum firing rate is decreased. According our previous study numerically, the maximum heat release rate for HiTAC is only 1/49 times for the same fuel capacity with a normal high velocity jet burner. The low maximum firing rate also implies that the combustion noise is possibly small since any approach that reduces combustion intensity within a combustion reaction may be expected to reduce the sound power produced by a flame. 7.1.5 Heat Transfer Elevation Heat transfer was evaluated in a test furnace equipped with the one-flame HRS has been performed, which include a stationary sink and a moving sink. Figure 7.8 summarizes some results. Various heat flux densities were obtained depending on the type of charge used. The highest values were obtained for the stationary heat sink. For the HiTAC mode with a stationary sink the value of the heat flux density was on average 162.9 kW/m2. For the flame mode it was in the range 91.4 kW/m2. This indicates that the heat flux density for HiTAC mode is 1.78 times that for the conventional flame mode on the same type of sink. The air cooling tubes characterize another distribution of the heat flux density. For the HiTAC mode the value of the heat flux density was in the range 36.8 to 40.8 kW/m2. For the flame mode it was in the range 21.0 to 27.2 kW/m2. A 59 % greater average heat flux density for Case 0 was demonstrated. The total radiation heat flux density for a stationary heat sink depends very much on the combustion mode. For the HiTAC mode with the stationary heat sink, the value of the radiation heat flux density was in the range 191.0 to 205.4 kW/m2. For the conventional mode it was in the range 108.7 to 117.6 kW/m2. The average difference of total radiation for these two firing modes is the same proportion as the net heat flux as shown above. 108 Total radiation heat flux density along the furnace wall depends also very much on the combustion mode. For the HiTAC mode the value of the heat flux density was around 200 kW/m2. This value is similar to the total radiation heat flux on the top of the stationary heat sink. For the conventional mode it was in the range 138 to 151 kW/m2. Net heat flux for sink, kW/m2 200 160 120 80 40 0 0 500 1000 1500 2000 2500 3000 Distance from burner face, mm Case 0 Case 2 Case 4 Case 1 Case 3 Figure 7.8 Predications of heat flux absorbed by the charge along central line on the surface of sink Case 0: Test furnace with a one-burner HRS without any charge or heat sink. Case 1: Test furnace with conventional turbulent jet flame without any charge or heat sink. Case 2: Test furnace with a one-burner HRS and with a stationary heat sink whose surface temperature is equal to 20OC and constant, Case 3: Test furnace with conventional turbulent jet flame with a stationary heat sink whose surface temperature is equal to 20OC and constant. Case 4: Test furnace with a one-burner HRS with a moving steel slab which initial surface temperature is equal to 20OC. 109 7.1.6 Simulation of a Moving Slab In preparation to calculate a real industrial furnace with a moving heat sink, a moving slab is assumed in HiTAC test furnace equipped with a one-burner HRS (High-cycleRegenerative burner System). The slabs’ initial surface temperature was equal to 20OC. The moving slab was assumed to be made of a low carbon steel in the form of a moving plate. The moving slab was treated as a charge and its heating was calculated for one side facing the in-furnace processes. The total heat transfer surface of the moving slab was equal to 2.945 m2 and the heating capacity of the furnace was 1.25t/h. The steel slab moves with velocity equal to 0.000833 m/s along the furnace length beginning from the furnace “inlet” located below the burner. The distribution of the surface temperature of the moving slab heated under the one-flame HRS are shown in Figure 7.9. The slab’s surface temperature increases gradually and is fairly uniform across the furnace width. It should be noticed that the slab end temperature is 608.5 K. This is due to the limited length of the test furnace and due to the initial surface temperature, which is equal to 20 oC. The test furnace can be treated as a short section of a real heating furnace. 110 Figure 7.9 Predicted temperature distribution of moving slab with one-flame HRS 7.2 Study of the HiTAC furnace with a twin-flame HiTAC burner 7.2.1 Experimental and verification of the modelling prediction. Experiments measurements and numerical simulations on this semi-industrial-furnace equipped with two sets of two-flame HRS have been performed. Further more, the numerical modelling results were validated by comparing with experimental results including in-furnace species and temperature. In this work, only the profiles of the compositions of CO and O2 in furnace are compared between predicted and measured. The predicted were carried out when burner B and D were fired, and the measured values are average data after several cycles of burner’s firing. Figure 7.10 shows the comparisons for the concentrations of CO by depth through the furnace chamber at x=0 for z=1.7m along y direction, at x=−0.475 m for z=1.4m (near burner B) along y direction, and at x=0 for z=0.75m along y direction, respectively. Generally, the overall agreement may be accepted considering the complexity of 111 fluctuations of turbulent combustion process in the furnace, the assumptions made in the numerical model and the uncertainty of the experimental data. The measured shows larger zone of CO existing than that of the predicted, the most possibility reason is only half cycle of firing was simulated. Figure 7.11 presents the measured and predicted concentrations of O2 by depth through the furnace chamber at x=−0.475 for z=1.4 m (near burner B) along y direction respectively. The predicted concentrations of O2 follow similar trend to measured values, exhibiting the same peak concentration location in Figure 4 (b), which is very close to the burner B. 112 5000 Predicted-CO Measured-CO 3000 2000 1000 0 0 100 200 300 400 500 600 700 Distance from furnace horizontal central plane, mm 800 (a) 5000 CO (ppm,dry) 4000 Predicted-CO Measured-CO 3000 2000 1000 0 0 100 200 300 400 500 600 700 Distance from furnace horizontal central plane, mm 800 (b) 400 CO (ppm,dry) CO (ppm,dry) 4000 Predicted-CO 300 Measured-CO 200 100 0 0 200 400 600 Distance from furnace horizontal central plane, mm 800 (c) Figure 7.10 Predicted and measured CO profiles in the upper part of the furnace (a) x=0, z=1.7m, (b) x=-0.475 m, z=1.4m, (c) x=0., z=0.75m 113 Concentration of O2,/%,dry) 12 Predited-O2 9 Mearsured 6 3 0 0 100 200 300 400 500 600 700 800 Distance from burner central surface Figure 7.11 Predicted and measured O2 profiles in the upper part of the furnace at x=-0.475 m, z=1.4m 7.2.2 Effect of flame configurations Figure 7.12 depicts the temperature distributions on the horizontal plane, including the central plane, of injection ports of combustion air for different flame configurations which used in this study are Counter mode, fired by combining of burners B and D, Single-side mode, fired by combining of burners C and D and Stagger mode, fired by combining of burners A and D. T, K (a) (b) (c) Figure 7.12 Temperature profiles (K) at various firing locations (a) Counter mode (b) Single-side mode, (c) Stagger mode 114 The peak temperatures of the gases are 1856, 1821 and 1900 K for combustion modes of counter, Single-side and stagger, respectively. The maximum flame temperature occurs at the case of stagger mode. Results from experimental study show the same conclusion. This is the fact that the hot gases recirculation formed by burners for stagger mode reduces the heat loss from flame. These maximum temperatures in the furnace at various modes are lower though the combustion air was preheated to 1311 K. This is a result of very intensive flue gas recirculation created by the internal recirculation zones. The highest temperature zones are found to occur in the middle of the combustion chamber at the burner level, farther away from the burner face. The counter mode gives the longest distance from burner face where the maximum temperature appears, while the stagger mode gives the shortest. Furthermore, because the flame from this burner is less visible (sometimes invisible) than the combustion flame from a conventional burner because of the oxidizer diluted by the internal recirculation, it is therefore generally accepted that flame length is not suitable parameter for characterizing flame size. Instead, it is necessary to demonstrate flame shape and size using a comprehensive numerical simulation. Figure 7.13 shows the influences of flame configurations on flame volumes. Because of difference of burner’s configurations, the flame shape and size are difference. 115 (a) (b) (c) Figure 7.13 Flame shapes shown by the oxidation mixture ratio RO=0.99 for difference firing modes (a) Counter mode (b) Single-side mode (c) Stagger mode 116 The flame volume and the furnace flame occupation coefficient are summarized in Table 1. The Counter flame occupies the smallest fraction of the furnace volume and the Single-side flame occupies the biggest fraction of the furnace volume. Consistent with this conclusion, the Counter mode has maximum local peak firing rate and maximum flame combustion intensity. This implies that the combustion noise in Counter mode is larger than in the other two modes since any approach that reduces combustion intensity within a combustion reaction may be expected to reduce the sound power produced by a flame. Figure 7.14 shows the influences of flame configurations on NO emissions. Generally, the NO emission is at low level for all studied cases. The differences in NO emissions can be clearly seen among these cases. The stagger mode has highest emissions of NO and the Single-side mode leads to lest emissions. The NO emission corresponds to the flame NO emission, mg/MJ maximum temperature. A higher the flame temperature leads to a higher the NO emission. 60 40 20 0 Counter Parallel Stagger Firing Configureation Figure 7.14 Effect of firing configurations on NO emissions 7.2.3 Effect of excess air ratio The internal recirculation of the flue gas is key issue to successful contemporary regenerator burner system. Therefore the momentum of the combustion air or fuel defining 117 this internal recirculation plays a very important role in this type of burner. For a given fuel flux, the excess air ratio plays a central role. The effects of the excess air ratio on NOx level have been investigated numerically in this study for excess air ratios from 1.04 to 1.45. In all of these simulations, the thermal input of the fuel to the burner has been kept constant. Other boundary conditions have been also kept constant during the simulation comparison in the test case used above. The increase of excess air ratio increases the velocity of combustion air. Figure 7.15 depicts the measured and predicted NO emission effected excess air ratio for the counter mode. The measured NO emission generally agreed with the predicted values, NO emission, mg/MJ and exhibited similar changes in ratios of excess air. 60 Predicted Experied 45 30 15 0 1,0 1,1 1,2 1,3 1,4 Excess Air Ratio for counter mode 1,5 Figure 7.15 NOx emission vs excess air ration during the combustion of counter mode Generally, NO emission increases with increasing excess air ratio and this trend of increase become weak after excess air ratio is higher than of 1.3. This is caused by the associated increase in oxygen availability and flame temperature. On one hand, the velocity of injected air increases with increase of excess air ratio, more combustion products are entrained into the root of the flame, thus the temperature of flame decreases. However, the velocity of increased injected air leads to better mixing, and maximum temperature of flame increases. On the other hand, combustion reaction rates are depressed because the fuel and air mixture is diluted by gases entrained from the furnace atmosphere. Therefore, chemical reaction 118 zone is not wholly confined by a burner tunnel or quarl. Consequence, the concentration of oxygen is quite sensitive to NO formation. This result also indicates that the production rate of NO is also controlled both by turbulentmixing and by chemical kinetics. When excess air ratio increases a slight from stoichiometric ratio, both the chemical kinetic (oxygen concentration) and turbulent-mixing contributes the NO product rate. If the excess air ratio increases further, NO product rate would be controlled major by turbulent-mixing. 7.2.4 Effect of Fuel/Air injection momentum ratio NO emission Previous we argued that the actual NOx emission is obscured by the ratio of the fuel and air injection momentum. In Figure 7.16, NOx emissions for various fuel/air injection momentum ratio were shown. It was found that the fuel/air momentum ratio has a strong influence on the NO emission in the furnace. Less the ratio of fuel/air momentum, larger the NO production, especially when the ratio of fuel and air momentum is low. The explanations for this trend have been shown in previous. Summary, the emissions of NO at this technology depend on the oxygen availability and flame temperature. A higher ratio of fuel/air injection momentum, for example, a higher momentum of fuel, or a lower momentum of air, decreases the mixing of fuel and air within the primary combustion zone. This leads to a lower maximum temperature and more uniform distribution of reactants inside the flame and larger chemical reaction region. If the changing of momentum of ratio is caused also by excess air ratio, availability of oxygen plays also very important role for the emissions of NO. 119 NO emission, mg/MJ 100 80 Predicted-Stagger Measured-Counter Predicted-Counter 60 40 y = 0,6204x -1,7991 20 0 0,05 0,15 0,25 0,35 Fuel/Air Injection Momentum Ratio Figure 7.16 NO variations with the ratio of the fuel/air injection momentum 7.3. OPTIMAL DESIGN OF A HiTAC FURNACE 7.3.1 Flame Entrainment Ratio Figure 7.17 presents the distribution of Rfe along flame length (excess air ratio equal 1.09). Here, the flame length is calculated as the distance between the end of fuel nozzle and the axial location of the oxidation mixture ratio equal to 0.99. When the normalized distance—defined by the ratio of the flame length to the burner diameter—equals 1, the entrainment is very small. Above 1, the entrainment increases with increasing of the flame length as show in Figure 7.17 with 1.09 of excess air ratio. The highest flame entrainment ratio 4.1 occurs at the distance of 1.41 m corresponding to half of the flame length approximately. At this cross section, the diameter of the flame is around 0.51m (the normalized flame diameter 2.32). After this section, the flame entrainment ratio almost keeps constant. 120 6,0 ALFA=1.09 5,0 ALFA=1.15 Entrainment Ratio ALFA=1.26 4,0 3,0 2,0 1,0 0,0 0 2 4 6 8 10 12 14 z/D Figure 7.17 Flame entrainment ratio vs. excess air ratio (ALFA) The internal recirculation of the flue gas is the key issue to a successful realizing the HiTAC technology. Therefore the momentum of the combustion air and the fuel that determine this internal recirculation plays a very important role. For a given fuel flow rate, the excess air ratio plays a central role. The effects of the excess air ratio on NOx level and flame volume have been investigated numerically, for the excess air ratios from 1.09 to 1.26. In all of these simulations, the fuel input to the burner has been kept constant as shown in Table 7.1. Other boundary conditions, for example the distribution of the cooling tube wall temperature, have been also kept constant. The influences of the excess air ratio on the flame volume, the flame entrainment ratio, Rfe, and NOx emissions have all been investigated. Figure 7.18 shows the furnace flame occupation coefficient, RFOC for various excess air ratios. The RFOC decreases with the increasing of the excess air ratio. For example, when the ratio of excess air is 1.09, a much larger flame volume appears; around 2.8 times larger than when λ =1.26. The explanation for this phenomenon is that the lower excess air ratio limits the mixing of the fuel with the combustion air in the primary combustion region, leading to a lower combustion rate. Thus, a larger combustion reaction zone is required. 121 2000 3,5 1900 RFOC, % 3,0 1800 2,5 1700 2,0 Temperature, K 4,0 1600 1,5 1,0 1500 1,09 1,15 Air Excess Ration 1,26 Flame-Occupation-Coefficient The maximum temperature Figure 7.18 Excess air ratio vs. flame occupation coefficient and peak temperature It is also apparent that the maximum temperature in the furnace is strongly affected by the excess air ratio as well. A larger excess air ratio leads to a larger maximum temperature (Figure 7.12). One of the possible reasons is that the oxygen concentration in reaction zone increase as excess air ratio increase, which resulting in a higher flame temperature. Another reason is a smaller flame entrainment at a larger excess air ratio leads a weaker dilution of the flue gas, thus a higher the flue gas’s peak temperature. Figure 7.17 presents the distribution of the flame entrainment ratio for different excess air ratio. Generally, when the excess air ratio increases, the flame entrainment decreases because of the smaller flame volume as interpreted previous. This also implies that a larger gross injection momentum does not mean a larger flame entrainment ratio. This is different from the conventional high velocity jet burner. Furthermore, the profile of the flame entrainment ratio is also affected by the excess air ratio. When the excess air ratio increases above 1.09, the entrainment ratio has a maximum value at the half of the flame length. Figure 7.19 shows the flame entrainment ratio as a function of the normalized distance, which is defined as the ratio of flame length to burner diameter, for HiTAC test furnace equipped with two sets of two-flame HRS fired with staggered and counter modes. 122 Flame entrainment Ratio 4 3 2 1 0 0 2 4 6 Normalize distance 8 10 Stag. Tf =293 K Stag. Tf =573K Stag. Tf =873 Counter Tf =293K Figure 7.19 Effect of fuel temperature and flame locations on flame entrainment ratio The effect of fuel temperature is obvious. When the fuel temperature is 293K, the location of the maximum flame entrainment is away from the flue gas outlet. In this case the CO emission is zero. When the fuel temperature is increased, the fuel velocity increases, the location of the highest flame entrainment is moved more closely to the outlet, and the combustion efficiency is decreased. According to the simulation, the emissions of CO are 118, 183 and 250 ppm (mass) for the fuel temperatures 573, 873 and 1273 K, respectively. Based on these results, an optimum furnace width can be designed. The counter mode offers a different profile of the flame entrainment ratio since the flows of two firing burners ru n into each other. The profile of the flame entrainment ratio is symmetrical (Fig.7.18). The value of entrainment increases with flame length from both burner faces. There are two peaks for the entrainment ratio and they occurr at about 4 burner diameters from both burner faces. The flame entrainment is very small at the middle of the two firing burners. This is because the flow direction of the flue gases in the flame is changed from a vertical burner face to single-side burner face as shown in Figure 7.12. 123 Generally, the highest flame entrainment ratio is found at the same normalized distance (around 3) although they occurred at different locations since the total momentum of the burners for the two modes studied was kept constant. This could be used to estimate an optimum furnace design. 7.3.2 Optimal Design of HiTAC Furnace The purpose of the design of a high performace industrial furnace is to provide a high and uniform temperature in the furnace and a low NOx emission without sacrifying the combustion efficiency. HiTAC has the characteristics of slow combustion diluted by internal flue gas recirculation. In order to maintain enough internal flue gas recirculation for a specific burner configuration, a minimum flow area in the furnace cross section is requried. Furthermore, the width of furnace (distance from burner face to exhaust exit) should also be maintained otherwise the problem of ejection of unburned fuel gases through the discharge burner arises. In this part of the work these two parameters are discussed. It is clearly seen from the above results and discussion that a high entrainment ratio of the flue gas should be maintained. Therefore, according the flame shape and the flame entrainment ratio, the miminun flowing area of recirculation zone and furnace length can be determined. The minimum cross section of the flue gas recirculation flow could be founded at the section where the maximum flame diameter exists. Figure 7.20 shows the scheme of the flame zone and the flue gas recirculation zone for a HiTAC furance equipped with any type of HRS. In this study, it is assumed that all the positive velocities across the cross section of the furnace are included in the flame section. This is a good assumption in the part of furnace from the burner face to the maximum flame cross-section. 124 mr mout mout-2 mf mo Figure 7.20 The scheme of structure of the flame zone and recirculation in an enclosed combustion chamber The mass balance for this part of the chamber extending from the burner face to any flame section is as follows: m0 + mr = m f + mout (7.1) where, mo is the mass flow rate including the fuel and the combustion air, mf is the mass flow rate through the flame cross-section, and mr is the mass flow rate of the flue gas recirculation and mout is the mass flow rate of the gases leaving the chamber. When Equ. 7.1 is divided by m0, one obtains: mr m f m0 − mout = − m 0 m0 m0 (7.2) The flow rate, m, flows any cross section can be calculated as: m = ρv A (7.3) Here, ρ , v and A are the flue gases average density, the velocity and area, respectively According to the definition of the flame entrainment ratio in Eq.3.29, and Equ.7.2, the area of flue gas recirculation can be calculated from Equ.7.4, expressed as: 125 m0 − mout ) m0 ρ r vr mo ( R fe − Ar = (7.4) The minimum cross section of the flue gas recirculation flow could exist at the section where the maximum flame entrainment ratio, Rfe. The critical area of the flue gas recirculation, Ar,cr can be determined by the maximum flame entrainment ratio Rfe. mo − mout ) mo ρ r vr mo ( R fe ,max − Ar ,cr = (7.5) It can be seen that the criteria minimum area of combustion chamber is function of: • the maximum flame entrainment rate of the burner, Rfe,max,, • the initial total mass flow rates of the burner, mo, • density of in-furnace flue gases, thus furnace temperature, • velocity of in-furnace flue gases which are effected by furnace temperature and the length of the furnace along the vertical direction of the burner’s face. • the type of the HiTAC system. i.e., one-flame or two-flame HiTAC system. According to the design of the studied burner system in this paper, mout = 0.8m0 (7.6) One obtains: Ar ,cr = mo ( R fe ,max − 0.2) ρ r vr (7.7) Here, Ar,cr is the minimum cross area needed for flue gas recirculation, mo is initial total mass flow rate of the burner. ρ and v are the density and the velocity respectively, r denotes the recirculated flue gas in the combustion chamber. 126 From the Eq.7.7, it is easy to calculate a minimum flowing area of the recirculation flue gas. For example, if this one-flame HiTAC system is installed on a cylindrical furnace, the maximum flame entrainment ratio is 4.1 at the half of the flame length and this flame diameter is 0.51 m. The temperature of the furnace is around 1373 K, and the average density is around 0.28. Additionally the average velocity of the recirculating flue gases is approach 1 m/s in the cross section of Rfe,max at this furnace temperature. Therefore, the minimum diameter of the chamber is 0.7 m. This is approximately 3.2 times of the burner diameter. According to this, it is easy to give the optimal design of HiTAC furnace. For example, for the stagger firing mode of the burner studied and for the design condition, the Rfe,max is 2.6 according to this calculation. The temperature of the furnace is around 1373 K, and the average flue gas density is around 0.28. Additionally the average velocity of the recirculating flue gases is approaching 1 m/s in the cross section of Rfe,max at this furnace temperature. The area of flame cross section at the Rfe,max is 0.155168 m2, which occurs at 1.2m from burner face. If this burner is placed in a furnace with a cylindrical combustion chamber, the critical area of flue gas recirculation is 0.04m2, therefore, the critical diameter is 0.5m. This is approximately 2.7 times of the burner diameter. Another parameter for optimal design is the width of the furnace. Figure 7.21 depicts the distribution of temperature, through the central point of the burner B and D, and flame entrainment ratio along the direction of the furnace width for HiTAC the test furnace with two-flame HRS burners in the staggered firing mode. The optimal design of the width of furnace depends on the optimal location of the peak flame entrainment ratio. 127 3 1800 2 1400 1 Flame temperature 1000 Flame entrainment 600 Flame entrainment ratio Temperature, K 2200 0 0 2 4 6 8 Normalize Distance 10 Figure 7.21 Distributions of Temperature and Flame entrainment ratio along the direction of the furnace width The following three aspects are considered to determine the optimum location of peak flame entrainment ratio. According to the characteristics of HiTAC, a large entrainment ratio leads to low NOx emissions. It therefore follows that if the peak temperature occurs at the location of the peak flame entrainment ratio, the lowest NOx emission will be realized for a specific type of burner. Therefore, the location of the peak flame entrainment ratio should be designed to exist as close to the firing burner as possible. Secondly, the maximal uniform temperature profile in the furnace can be achieved when the peak flame entrainment ratio is located at the middle of the furnace width. Finally, the combustion efficiency in the flame increases with the increase in the flame width. In order to achieve complete combustion, a sufficiently large furnace width is necessary. To summarize, the maximum flame cross section should be at least half of the furnace length. Consequently, for a HiTAC furnace fitted with two-flame HRS, the optimum furnace length is in the range of 1.2−2.4m for the burner studied in a staggered firing configuration. This is approximately 6.5−13 times the burner diameter. Similarly, the optimum design of the HiTAC furnace equipped with a one-flame HRS can also can be determined. It is assumed that the one-flame HRS is installed on a cylinder combustion chamber, and the temperature of the furnace is 1373 K, the average flue gas 128 density is around 0.28 and the average velocity of recirculation flue gas is approaching 1 m/s at the cross section of Rfe,max at this furnace temperature. For the above design, the maximum flame diameter is 0.5m and exists at the half of the flame length and the entrainment ratio is equal to 4.0. Therefore, the minimum diameter of the chamber is 0.7 m. This is approximately 3.2 times the burner diameter. It can be seen for the same fuel capacity, for example, 200 kW, that the optimum width of combustion chamber equipped with two-flame HRS is 1.0 m, this value for one-flame HRS is 0.7 m. Therefore it can be concluded that the optimum width of a combustion chamber equipped with a two-flame HRSis larger than that for a one-flame HRS. 7. 4 CONCLUSIONS A HiTAC test furnace equipped with HRS (both one-flame and two-flame) was studied experimentally and numerically and the main conclusions are as follows: 1. For the combustion model, the Eddy-Dissipation-Concept with multi-step chemical reactions is a more suitable numerical model for HiTAC, especially when modelling is applied to large scale industrial furnaces. 2. The concepts, including oxidation mixture ratio, furnace-gas-temperature-uniformityratio, Furnace Flame Occupation Coefficient and Flame entrainment ratio, were defined to describe the characteristics of HiTAC, and are used to optimize the design of a HiTAC furnace and burner. 3. The benefits of HRS are quantitatively demonstrated by mathematical models. They are: lower peak temperature, larger flame volume, a more uniform thermal field, lower local firing rate, higher heat transfer, higher energy utilizing efficiency and lower combustion noise. 4. Operation parameters, including oxygen concentration, combustion air temperature, fuel temperature, fuel flow rate, excess air ratio and flame locations have strong influences 129 on the combustion and NO emissions in the HiTAC furnace. The optimum combination of these parameters should be considered 5. The criteria for the choice of the diameter and length of a furnace fitted with HRS burners are proposed in order to achieve an optimum design for HiTAC operation. 130 8 NO formation and destruction Mechanism study For HiTAC, the large quantities of recirculation of combustion products are entrained into the fresh reactants before combustion thus higher peak temperature (T<1850 K) is lack. As a result, thermal NOx is suppressed and much of NO maybe form mainly by mechanisms that are insignificant in most conventional combustors. One of most possibility routines is the NO formation via nitrous oxide mechanism (N2O). In the present work, N2O-intermediate NO model is developed. This model is accompanied with exiting thermal, prompt, and reburning NO models for prediction of the NO formation and emission in a semi-industrial furnace equipped with a HiTAC burner. The sensitivity of furnace temperature and oxygen availability on NO generation rate has been also investigated. The predicted results were compared and analysis with experimental values. The numerical simulations on this semi-industrial furnace equipped with one-flame HiTAC system have been validated in our previous study by comparing with experimental results including energy balance, in-furnace species and temperature. The agreements of predicted and measured are very encouraging, especially the numerical simulation shows a very good prediction of the HiTAC flame’s shape and size. In the present study, only NO models were concentrated. 8.1 Experimental and verification of NO emissions models Figure 8.1 shows the results from experimental measurement and numerical prediction with and without N2O route at different excess air ratio. The measured values were obtained at the average of burner outlet and chimney of the furnace. The estimated uncertainties of the measured NOx values were 95 percent confidence level. The experimental data shown were the values of NOx (NO+NO2) and the numerical calculations are only for NO. However, the gradual oxidation of a small amount of NO to NO2 in the immediate post flame gases does not alter the total NOx emission significantly. Additionally, in order to eliminate the dilution effect by excess air, the NO concentrations are corrected to mg/MJ fuel. 131 NO, mg/MJ 80 Predicted without N2O route Mesured Predicted with N2O route 60 40 20 0 1,0 1,1 1,2 1,3 Air Excess Ratio Figure 8.1 Effects of excess air ratio on NO emission Generally, it can be seen that NO emissions are lower than the limit of emission standards for industrial furnace (100mg/MJ) although combustion air was preheated 1207K. The numerical calculated NO emission without N2O-route model is lower than that obtained by experimental measurement of NOx emissions. This difference is clearer at the cases of lower excess air ratios. For example, the NO model without N2O route could almost not predict NO emissions in the case of excess air ratio equal to 1.04 because of the predicted value is almost zero. Although only NO emission was predicted, these differences are much larger than errors led by ignoring of reading of NO2, which is only in order of 10% during experiments. When N2O route was added to predict NO emission, the calculated NO emission showed a significant improvement. For instance, the NO emission obtained based on the models with N2O route is around 2.7 times larger than that predicted without N2O route in the case of excess air ratio equal to 1.09. At 1.15 of excess air ratio, this relation is 2.4. The approximate percentages of NOx production by nitrous and other three mechanisms (thermal, prompt and reburning) are 99:1, 73:27 and 70:30 at excess air ratios equal to 1.04, 1.09 and 1.15, respectively. This result implies that NO emission via N2O intermediate is significant and for lower excess air ratio this model will gain more importance. This conclusion is consistent with the model results in the work [33] in the case of methane combustion in a stirred reactor. 132 Comparing with experimental data, the results of the NO emission predicted with N2O route is very encouraging when the excess air ratio is equal or less than 1.15. However, when excess air ratio is equal to 1.25, predicted NO emission with N2O-route was 52 mg/MJ, which is almost 2 times higher than measurement. In this case, proposed model of NO emissions via N2O-mechanism fails to predict NO emission, existing NO models (thermal, prompt and reburning NO) are enough exactly to predict NO emissions as shown in Figure 8.1. In order to further understand this result, the variation of the maximum furnace temperature is given in Figure 8.2 as a function of excess air ratio. It can be seen that the peak temperature in the furnace rises as excess air ratio increases. This is the fact that the oxygen concentration in reaction zone increase as excess air ratio increase, which resulting in a higher flame temperature. When excess air ratio is lower than 1.15, the peak temperature is lower than 1850 K. When excess air ratio is raised to 1.25, the peak temperature is over 1900 K. Max. T, K 1950 1850 1750 1650 1,0 1,1 1,2 1,3 Excess Air Ratio Figure 8.2 Effect of excess air ratio on maximum temperature Additionally, Figure 8.2 also clearly shows that the NO concentration is very sensitive to the excess air ratio and increases linearly with excess. This is caused by the associated 133 increase in oxygen availability and flame temperature. On one hand, a larger excess air ratio leads to increase of oxygen availability, consequently, increase formation of NO via N2O+O. On the other hand, the velocity of increased injected air leads to better mixing, and maximum temperature of flame increases, thus NO formation rate increase since NO formation rate depends on temperature. As reported by Sarofim and Flagan [119] that the rate of thermal-NOx formation is significant only at temperatures greater than 1823 K and doubles for every increase in temperature of about 40 K. Therefore, when flame temperature is lower than 1850K, NO emission rate via N2O-route plays a very important role. When flame temperature is higher than 1850K, NO formation rate via thermal-NO is of importance. Present thermal-NO model woks well because of its self-compensating (big errors in measurements and derived reaction rates [120]). Therefore, developed NO model via N2O-machanis have to be further improved in order to obtain high accuracy prediction of NO emissions. This will be further argued in the next part. 8.2 Effects of NO model on the NO formations In order to further exam effects of the NO model on NO formation and destruction in the test furnace, it is very useful to investigate the temperature profiles as shown in Figure 8.5, which at the cross-section through the fuel and one air nozzle in the test furnace at different excess air ratio. Generally, there is a very large chemical reaction zone in all cases. This is also verified by experimental studies. The highest temperature zones are found to occur in the middle of the combustion chamber at the burner level, farther away from the burner face. HiTAC burner used in this study limit the mixing of the fuel with the combustion air at the initial stage of the combustion. Large combustion products are entrained into the root of the flame because of the high injection momentum of it’s nozzles, this reduce the oxygen availability in the primary combustion zone. The fuel that is not burned in this region gradually mixes with air to complete the combustion. This results in a more uniform temperature profile, which corresponds to a lower NO production rate. 134 Predicted NO contours at the cross-section through the fuel and one air nozzle in the test furnace at different excess air ratio (λ) with and without N2O-route were shown in Figure 8.3 and Figure 8.4. It can be seen that the predicted NO formations with and without N2Ointermediate mechanism are quite difference. Firstly, the locations of maximum NO formation are different. For example, when N2O-intermediate mechanism has not been included in the cases of excess air ratios equal to 1.09 and 1.15, these locations are far from the burner zone, which is consistent with where the peak temperature appears. This implies that temperature plays the major role for the formation of NO. However, when N2Ointermediate mechanism was considered, the location of the maximum NO formation is closer to the burner zone. In particular, these locations are occurred near the nozzles of fuel and air inlet when excess air ratios equal to 1.09, 1.15 and 1.25. This implies that NO formation and destruction are strong affected by the radical species as well as flame’s temperature. 135 (a) without N2O-route (b) with N2O-route at at λ= 1.04 λ= 1.04 (c) without N2O-route (d) with N2O-route at at λ= 1.09 λ= 1.09 Figure 8.3 Predicted NO contours (mass%) at the cross-section through fuel and one air nozzle in the test furnace at excess air ratio (λ) equal to 1.04 and 1.09 with and without N2O-route 136 (e) without N2O-route (f) with N2O-route at λ= at λ= 1.15 1.15 (g) without N2O-route (h) with N2O-route at at λ= 1.25 λ= 1.25 Figure 8.4 Predicted NO contours (mass%) at the cross-section through fuel and one air nozzle in the test furnace at excess air ratio (λ) equal to 1.15 and 1.25with and without N2O-route 137 (a) λ=1.04 (b) λ=1.09 (c) λ=1.15 (d) λ=1.25 Figure 8.5 Temperature distribution at the cross-section through fuel and one air nozzle in the test furnace at different excess air ratio (λ) Secondly, the distributions of NO formation with N2O-intermediate mechanism is limited in the major chemical reaction zone as shown in our previous works, which the flame volume was figured out by experimental and numerical calculating studies. 138 Basing on these analyses, it could be concluded that the prediction of NO formation and destruction with N2O-intermediate mechanism is more reasonable. It is stipulated that for the case of excess air ratio equal to 1.25, the prediction of NO formation and destruction with N2O-intermediate mechanism is also reasonable. This implies that developed N2O-intermediate mechanism is able to predict the profiles of NO formation and destruction in high temperature air combustion, but it gives an over prediction of NO emission. Therefore, one of improvement solution is considering the limit of temperature. Therefore, the equation (3.13) can be modified as following: [ R N 2O _ NO _ i ⎧− min REBU _ NO _ i , RKIN ⎪⎪ =⎨ ⎪0 ⎪⎩ ] T < 1850 K T ≥ 1850 K kg/m3s (8.1) This improvement was used to predict NO emission at the case of excess air ratio equal to 1.25. The predicted NO emissions were 35 mg/MJ, and the NO profile in the furnace was similar as before. This improvement is effective. 8.3 NO emissions From Coflow Gas Jet Combustion Study In this study here, the following NO formation and destruction models were used: • Thermal NO formation • Prompt NOx route • N2O route • NO-reburning The numerical results show that the N2O-intermediate mechanism is very importance during HiTAC operation at low preheated temperatures. For example, Figure 8.6 presents 139 the influences of oxygen concentration on NO emissions at the case of the co-flow gas jet in the combustion test case. It shows that when the oxygen concentration is larger than 10%, thermal and prompt NO emissions play a dominating role. However, when the oxygen concentration is less than 10%, the NO emissions formed from the nitrous oxide mechanism (N2O) have a substantial influence. For LPG combustion with a 1173K air preheat temperature, the approximate percentage of NO production by the nitrous oxide to Zeldovich and prompt mechanism vary from 5:95 at 10% oxygen concentration to 95:5 at 5% oxygen concentration. Generally, NO emissions decrease with reduction in oxygen concentration because of the lack of a peak temperature in the flame. 140 2,0 3000 1,5 2000 1,0 1000 0,5 0 0,0 0 5 10 15 20 ratio of NO via N2O and total NO emission, mg/MJ 4000 25 Oxygen concentration in combsution air,mass% Thermal+Prompt-NO N2O-path-NO N2O-path-NO/NO Figure 8.6 Effect of oxygen concentration on NO emissions at an air temperature of 1173K and fuel temperature of 299K for gas jet combustion in the co-flow test Furthermore, the increase in the preheat temperature of air leads to an increase of the NO production. However, this increasing trend is weak at low preheat temperatures. Figure 8.7 gives the variation of the NO emission with the preheat air temperature. A lower maximum temperature in the flame and NO formation from the N2O-intermediate mechanism can explain this very well. 141 NO emission, mg/MJ 3000 2500 2000 1500 1000 500 0 1000 1200 1400 1600 Preheated air temperature, K Figure 8.7 Effect of preheated air temperature on NO emissions at a fuel temperature of 299K for the case of gas jet combustion in the co-flow test The influence of fuel flow rate on the NO emissions are also considered. Generally, an increase in the fuel flowrate increases the NO emissions as shown in Figure 8.8. Figure 8.8 also indicates that this increase is much smaller at a lower oxygen concentration than at a higher oxygen concentration. NO emission,mg/MJ 2500 2000 1500 1000 500 0 0,E+00 5,E-06 1,E-05 2,E-05 2,E-05 Fuel flow rate QF,kg/s Oxygen=15,4%(vol.) Oxygen=9,4%,(vol) Figure 8.8 Effect of fuel flowrate on NO emission at an air temperature of 1173K and a fuel temperature of 299K for gas jet combustion in the co-flow test 142 8.4 Conclusions N2O-route is developed to predict NO formation and emission during high temperature air combustion because of its lack of peak temperature. This model was used to simulate NO formation and emission for a semi-industrial furnace equipped with HiTAC burner. The predicted results were compared with experimental values. The results show that NO emission formed by N2O-intermediate mechanism is of outstanding importance during low peak temperature. It can give more reasonable predicted profile of NO formation and destruction. Developed model provides a basis for further studies. Furthermore, increasing excess air ratio leads to increasing of NO emission. 143 9 Thermodynamics investigation of the ‘Flameless’ combustion The available energy of fuel Eav is the difference between Gibb’s energy of reactant at ambient condition and that of burned produce as the same condition: E av ≡ Ex I = G I − G E (9.1) Exergy, Ex and Gibb’s free energy, G, are expressed by using temperature, T, enthalpy, H,, and entropy, S: Ex = H − H 0 − T0 ( S − S 0 ) (9.2) G = H − TS (9.3) Where, subscript 0 denotes the ambient condition. The thermodynamic analysis is usefulness to describe quantitatively the regime of the flameless combustion. Here, only the H-T diagram is used. In Figure 9.1, the HR and Hp curves are the relation of enthalpy of unburned mixture and its preheat temperature and the relation of enthalpy of chemically equilibrated burned product and its temperatures, respectively. It can be seen that the HR is increased linearly with the increasing of the preheated temperature. The difference of the enthalpy of unburned mixture at different oxygen concentrations in the reactants is not so larger since it dependence strong on the temperature. However, the Hp-curve increases exponentially in the high temperature range due to the increase of apparent specific heat caused by the thermal dissociation in product. The effect of the oxygen concentration on the enthalpy difference is significant, and the pure oxygen combustion has the biggest enthalpy difference. 144 The maximum temperature rise in the combustion process of specific condition is an isenthalpic combustion. This temperature rise is decreased obviously with the oxygen concentration in the mixture, and the increasing of the reactant temperature. It becomes zero at the cross point of the two curves, which corresponds to the adiabatic limit temperature. Ideally, the combustion without temperature rise can be realised by an isothermal combustion. The enthalpy difference at the isothermal combustion implies the maximum available energy. It can be seen that the oxygen-combustion has the maximum available energy. 12000 Taut Enthalpy, KJ/kg 8000 Tmax-NOx 4000 0 -4000 -8000 -12000 0 1000 2000 Temperature, K 3000 4000 HR-CH4/O2 Hp-CH4/O2 HR-CH4/21%O2 Hp-CH4/21%O2 HR-CH4/5%O2 Hp-CH4/5%O2 Figure 9.1 Enthalpy-temperature diagram for CH4 mixture with different oxygen level at the stoichiometric In order to clear demonstrate the effect of the oxygen concentration and the reactants’ temperature on the temperature rise, Figure 9.2 shows the relationship of the excess enthalpy and inlet temperature at different oxygen concentration for a methane/oxygen/nitrogen mixture. Here the excess enthalpy, ∆h, is calculated as: 145 Δh = Tad − Tin Tin − Tref (9.4) Here, Tad is the adiabatic temperature (K) of the reactant mixture with respect to inlet temperature Tin (K) of the reactant mixture, K. Tref is reference temperature, taken 273K, for example in this work. Figure 9.2 shows that the lower excess enthalpy during the combustion process can be achieved at a higher reactant temperature for the combustion with all the oxygen concentrations range. Excess Enthaply in log( (Tad-Tin)/(Tin-Tref)) 100,0 Nonisothermal Combustion 10,0 100 % oxygen 1,0 Quasiisothermal Combustion 21% oxygen 5% oxygen 0,1 0 1000 2000 3000 Temperature of the Reactant, K Figure 9.2 Excess enthalpy in log versus inlet temperature at different oxygen concentration for a methane/oxygen/nitrogen mixture Therefore, it is easy to find the ways to get a low temperature rise in the combustion process, for example, the followings may be taken: • preheating unburned mixture, • decreasing oxygen concentration (or lean combustion) in the unburned mixture, or • increasing the heat loss from the flame, or flame cooling 146 Preheating unburned mixture is very effective way to increase the energy utilization for the industrial furnace, where a larger enthalpy of the flue gas exists. This technology has been well studied. In fact, this is also effective solution to improve temperature uniformity. If the second way is used, the combustion stability should be considered. A preheated temperature is higher than the fuel’s autoignition temperature is preferred, which is widely used in the application of the furnace. We can also expect that any other ways to increase fuel flammability limitation technology might be used as well, for example, the catalytic combustion. The flame cooling is another technology to improve the combustion temperature uniformity. There are many different techniques which could be classified as flame cooling. A flame cooling tube might be also used to increase flame heat loss in order to get a low temperature rise during the combustion processor. The porous combustion is another example, where the heat released in the oxidative region is transferred to the unreacted mixture passing through a conduction channel inside the solid material or a radiative one inside the pores. Direct Flame Impingement (DFI) can be also classified this catalogue. Nitric oxide could be ranked to the most relevant pollutant when the combustion process is employed in the industries. Thermal NOx is favour when the temperature is higher than 1800K. It is well knowledge that the most effective way to depress NOx production is low the flame temperature for the high-temperature industrial application of fossil fuel combustion technologies. Obviously, the HiTAC, or MILD, or Flameless combustion concept is a combination of the first two items (and the third). The major characteristics of this available knowledge are: • The reactants must exceed self-ignition temperature, • This process evolves in a rather narrow temperature range especially when NOx emission is considered, and • The available energy is smaller, which leads to a low combustion intensity. It is generally agreed that the major characteristics of the HiTAC (or MILD) combustion is led by the low temperature increase in the combustion process. This is the fact that a low temperature increase makes a combustion chamber more similar as a well-stirred-reactor. 147 Again, the effect of heat release on the combustion characteristic is less at the case of a low temperature rise. For example, a larger flame entrainment is found during the HiTAC condition in or previous work. It is possible to liberate the above three limitations if only the mild temperature rise during the combustion process is considered. For example, the use of oxyfuel combustion can increase the combustion intensity and maintain a mild temperature rise as studied in this work. Basing on the above discussion, this new combustion phenomenon might be named as a Quasi-Isothermal Combustion, or QIC. It is obviously that the maximum allowable excess enthalpy of the QIC is decided by the fuel and oxidizer’s physical properties, and the combustion process, such as inlet, operations parameters. Additionally, the combustion stability should be maintained. The temperature of the reactants is higher that the fuel auto ignition is preferred. Generally, for the QIC, the maximum allowable excess enthalpy of the reactant mixture with respect to inlet temperature during combustion may be decided by the maximum excess enthalpy of the oxyfuel combustion (100% oxygen concentration) with respect to its self-ignition temperature. For methane, this value is 2.93 as shown in Figure 9.2. The studied burner in this paper is also one example of the QIC. It is realised by decreasing oxygen concentration before the combustion occur, and by flame cooling through the fluegas internal recirculation. 148 10 Conclusion The investigation has been performed using a single fuel jet flame facility with cross and co-flowing, and a semi-industrial furnace equipped with HiTAC burners. Experimental, numerical and theoretical analyzing investigations are adopted. In this work, a ‘chemical’ flame volume and ‘chemical’ flame length were used to describe this ‘invisible flame’. Results from single jet flame study show that: • Flame length increases with either the decrease of oxygen content, or increase of oxidizer temperature, or decrease of fuel temperature. Furthermore, the flame length is independent of the fuel flow rate and the diameter of the fuel nozzle for the studied cases. • Flame volume increases either with the decrease of oxygen content and increase of oxidizer temperature, or with the reduction of fuel temperature, or with the increasing in fuel firing rate. Flame volume depends very much on the oxygen concentration in the oxidizer. • Influences of high temperature and low oxygen concentration in the oxidizer on the flame Froude number, Frf were examined. Regimes of momentum- or buoyancy-control, were determined on the assumption that oxidizer temperature and oxygen concentration are changeable. A simple correlation of the ‘flame’ length and volume with flow parameters has been derived in terms of a flame Froude number for momentumbuoyancy transition jet flame under the HiTAC condition. The criteria constants of the dimensionless flame volume V* and the dimensionless flame length L* to assess momentum– or buoyancy–control flame are given. Additionally, the entrainments of this ‘invisible’ flame have been numerically and theoretical studied. Conclusions are: • The uniformity of the heat release in reacting jets has strong effect on the flame entrainment. More uniform the heat release, larger the entrainment. The effect of heat release reduces the entrainment in the near field of the reacting jets with the same factor of the characteristic ratio (Tf/To)0.5. 149 • The entrainment increases as the oxygen concentration is decreased. Furthermore, the entrainment is independent of the fuel flow rate and the preheated temperature of the oxidizer for the investigated temperature range (1073-1573K). • The effect of the oxygen concentration and preheated temperature of the oxidizer on buoyancy was examined. A correction Richardson coordinate, where the effect of the oxygen concentration (stoichiometric ratio) is included, was derived to describe the local influence of buoyancy force along the chemical flame length under the high temperature and oxygen deficient oxidizer condition. It can be concluded that the buoyancy force increases with the reduction of the oxygen concentration in the oxidizer. • The global behaviour of the entrainment was revealed. The entrainment of jet flames can be identified as two regimes: (a) the near field where entrainment coefficient is positive; and (b) the far field where entrainment coefficient is negative. Corrections of entrainment rates were derived in terms of a Frf number for momentum-buoyancy transition jet flame under the high temperature and low oxygen concentration oxidizer condition. Furthermore, the maximum entrainments along the flame length are estimated Further on, the benefits of HiTAC technology are quantitatively demonstrated by mathematical models. These benefits are: lower peak temperature, larger flame volume, more uniform thermal field, lower local firing rate, higher heat transfer, higher energy utilizing efficiency and lower combustion noise. NOx formation and destruction during this new combustion phenomenon has been studied numerically. It was found that the NO formation via N2O mechanism may be important. The approximate percentage of NO production by the nitrous oxide according to the Zeldovich and prompt mechanism varies from 5:95 at 10% oxygen concentration to 95:5 at 5% oxygen concentration. Finally, a thermodynamic analysis of combustion process shows that oxyfuel combustion is able to increases the available energy of the flameless combustion, thus higher combustion intensity. Additionally, the flameless oxyfuel burner does not need preheating the oxidizer, this extend the concept of the HiTAC. 150 Reference 1. http://www.iea.org/ 2. Tsuji H., Gupta A., Hasegawa T., Katsuki M., Kishimoto K., Morita M., High Temperature Air Combustion; From Energy Conservation to Pollution Reduction, ISBN 0-8493-1036-9, CRC Press LLC, New York, 2003. 3. Yasuda T, Ueno C., Proceedings of 2nd International Seminar on High Temperature Combustion in Industrial Furnaces, Stockholm, January 17-18, 2000. 4. Suzukawa Y, Sugiyama S, Hino Y, Ishioka M, Mori I. Energy Convers. Mgmt., 1997:38(10-13): 1061-1071. 5. Mori K. Outline of R&D and Field Test Demonstration Projects of High Performance Industrial Furnace, Proceedings of the Forum on High Temperature Air Combustion Technology, Josui Kaikan, Tokyo, October 11, 2001. 6. J. von Schéele, W. Blasiak, Oxyfuel solutions for lowered environmental impact of steel production, International Conference on Clean Technologies in the Steel Industry, 6-8 June, 2005, Balatonfüred, Hungary 7. W. Yang, W. Blasiak, T. Ekman, Numerical and thermodynamic study on oxyfuel flameless combustion, 8th International Conference on Energy and Clean Environment, Clean Air 2005, Lisbon, 27-30 June, 2005 8. K. Narayanan, W. Yang., T. Ekman, W. Blasiak, Combustion characteristics of flameless oxyfuel and oxygen-enriched-High Temperature Air Combustion, 6th HiTACG 2005, Sixth International Symposium on High Temperature Air Combustion and Gasification, 17-19 October 2005, Essen, Germany. 9. J. von Schéele, P. Vesterberg., O. Ritzen., Invisible Flames for Clearly Visible Results, Nordic Steel & Mining Review, pp 16 -18, 2005. 10. http://www.lindegas.com/international/web/lg/com/likelgcomn.nsf/docbyalias/ind_rebox 11. S.A. Lloyd and F.J. Weiberg, Natural, 251:47, 1974 12. Cornforth, J.R., Combustion Engineering and gas utilization, 3rd edition, 1992. 13. Tanaka R., New progress of energy saving technology toward the 21st century; Frontier of combustion and heat transfer technology; Proceedings of 11st IFRF members conference; 10-12th, May, 1995. 151 14. Hasegawa T, Tanaka R, Niioka T, Combustion with high temperature low oxygen air in regenerative burners. The first Asia-Pacific conference on Combustion, May 12-15, 1997, Osaka, Japan, p. 290-293 15. Hasegawa, T., Mochida, S., Gupta, A.K., AIAA Journal of Propulsion and Power, 2002:18(2):233-239. 16. Wünning, J. A, and Wünning, J. G. Progr. Energy Combust. Science, 1997:23(12):81-94. 17. A.K. Gupta A.K., S. Bolz, T. Hasegawa, Effect of air preheat temperature and oxygen concentration on flame structure and emission, Journal of Energy Resources Technology, September 1999, Vol. 121, pp 209-216. 18. Dally B.B., Karpetis A.N., Barlow R.S., Structure of turbulent nonpremixed jet flames in a diluted hot coflow, Proceedings of Combustion Institute, 29(2002)1147-1154. 19. Cavaliere A., De Joannon A., Mild Combustion, Progress in Energy and Combustion Science 30 (2004) 329-366. 20. Milani A, and Saponaro A., Diluted combustion technologies, IFRF Combustion Journal, article number 200101, February 2001, ISSN 1562-479X. 21. Blasiak W., and Yang Y, Narayanan K., and Scheele J von, Flameless oxyfuel combustion for fuel consumption and nitrogen oxides emissions reductions and productivity increase, 2007, vol 80, No.1. pp1--9. 22. Shimo N., Fundamental research of oil combustion with highly preheated air. Proceedings o the 2nd International seminar on high temperature combustion in industrial furnaces, January 17-18., 2000, Jernkontoret-KTH, Stockholm, Sweden. 23. Weber R., Orsino. S., Verlaan A.L:, and Lallemant N., Combustion of ligh and heavy fuel oils in high temperature air, Journal of the Institute of Energy, June, 2001,74,pp.38-47. 24. Blasiak, W, Yang W., Swiderski Artur, Kakietek Slawek, Study of Pulverized Coal Combustion in Rotary Kiln Using High Temperature Air to Minimize NOx Emission. Internal report, 2007. Royal Institute of technology, Sweden. 25. C. Lucas, D. Szewczyk., W. Blasiak, S. Mochida, Biomass and Bioenergy, 29-6 (2004) 563. 26. L. Yougn. C.C. P. Pian, Energy 28 (2003) 655. 27. K.Yoshikawa, Proceedings of the 19th annual international Pittsburgh coal conferences, Pittsburgh, PA; September 2002, Paper No. 31-2. 152 28. A. Ponzio, S. Kalisz, W. Blasiak, Fuel Processing Technology 87 (2006) 223-233. 29. Proceeding of Flameless Combustion for gas turbine, , in Lund, Sweden, June 21, 2005. 30. Kishimoto, K., Watanabe, Y., Kasahara, M., Hasegawa, T., Tanaka, R., observational study of chemiluminescence from flames with preheated and low oxygen air, The first Asia-Pacific conference on Combustion , Osaka, Japan, May 12-15,1997,pp468-471. 31. Amagai, K., Arai, M., Propane flame appeared in a hot air heated over selfignition temperature, RAN95, International Symposium on Advanced Energy Conversion Systems and Related Technologies, 4- 6 December, 1995 Nagoya, pp. 703-710 32. Kitagawa K., Konishi N., Arai N., Gupta A.K., Two-dimensional distribution of flame fluctuation during highly preheated air combustion, FACT-Vol.22, International Joint Power Generation Conference, Vol. 1, ASME 1998, p. 239242. 33. Isiguro T., Tsuge S., Furuhata T., Kitagawa K., Aral N., Hasegawa T., Tanaka R. and Gupta A. K.. Homogenization and Stabilization during Combustion of Hydrocarbons with Preheated Air. Proceedings of the Combustion Institute, Vol.27, 1998, 3205-3213 34. Plessing T., Peters N. and Wunning J. G. Laseroptical Investigation of Highly Preheated Combustion with Strong Exhaust Gas Recirculation. Proceedings of the Combustion Institute, Vol.27 ,1998, 3197-3204 35. Gupta A.K.,Li Z., Effect of fuel property on the struction of highly preheated air flames, International joint power generation conference, IJPGC 1997, ASME ECvol 5, Volume 1, 1997. 36. Bolz S, Gupta A. K., Effect of air preheated temperature and oxygen concentration on flame structure and emission, FACT-Vol.22, International Joint Power Generation Conference, Vol. 1, ASME 1998, p. 193-205. 37. Gupta A.K., Bolz S, Hasegawa T., Effect of air preheat temperature and oxygen concentration on flame structure and emission, Journal of Energy Resources Technology, September 1999, Vol. 121, pp 209-216. 38. Mochida S and Hasegawa T. Development of a Combustion Diagnostics Method on Advanced Industrial Furnaces Utilizing High Temperature Air Combustion. Proceedings of the 2 nd International Seminar on High Temperature Combustion 153 in Industrial Furnaces. January 17-18, 2000. Jernkontoret-KTH, Stockholm, Sweden.. 39. Lille S., Dobski T., Blasiak W., Visualisation of fuel jet in conditions of highly preheated air combustion, AIAA Journal of Propulsion and Power, 2000:16(4), p. 595-600.. 40. Lille S., Blasiak W., Mörtberg, M., Dobski, T., Yang W, Heat flux evaluation in a test furnace equipped with high temperature air combustion (HTAC) technology. Proceedings of IJPGC.02 2002 International Joint Power Generation Conference, Phoenix, AZ, USA, June 24-26, 2002. 41. Fujimori T., Riechelmann D., Sato J, Experiment study of NOx reduction by lifted turbulent jet flame in highly preheated flow, 1st ASPAC, Osaka, Japan, May 1997. 42. Fujimori T., Riechelmann D., Sato J.; Effect of liftoff on NOx emission of turbulent jet flame in high temperature coflowing air; 27th Int. Symposium on Combustion;Colorado, Boulder;1998. 43. Sato J, Combustion in high temperature air, 1st ASPAC, Osaka, Japan, May 1215, 1997. 44. Blasiak W., Szewczyk D., Dobski T., Influence of N2 addition on combustion of single jet of methane in highly preheated air, Proceedings of IJPGC’01, paper FACT-19048, New Orleans, USA, 4-7 June 2001. 45. Rota R., Guarneri F., Gelosa D., Effuggi A., Rabaioli M., Mild combustion in a small-scale aooaratus, 4th HTACG symposium, Rome, November, 2001. 46. Mochida S., Hasegawa T., Tanaka R., Characteristics of high temperature air combustion , RAN95, International symposium on advance energy conversion systems and related technologies, Nogoya, Japan, 4-6 December,1995. 47. Yuan J, Kobayashi Y., Naruse I, Combustion and NOx emission behaviors in regenerative combustion, 2nd International High Temperature Combustion Symposium, kaohsiung, Taiwan, January 20-22,1999. 48. Dong W ,Wlodzimerz Blasiak Largr eddy simulation of a single jet flow in highly preheated and dilute air combustion, Archivum combustionis Vol.20 (2000)No.3-4. 49. Dong W., Design of advanced industrial furnaces using numerical modeling method, Doctoral thesis, Royal Institute of Technology, Stockholm, Sweden. 2000 154 50. Yang W. and Blasiak, W., Numerical study of fuel temperature influence on single gas jet combustion in highly preheated and oxygen depleted air. The 3rd International Symposium on Advanced Energy Conversion Systems and Related Technologies, Dec.15-17, 2001, Nagoya, Japan. 51. Mörtberg Magnus, Study of gas fuel jet burning in low oxygen content and high temperature oxidizer, Doctor thesis Royal Institute of Technology, Sweden, 2005. ISBN 91-7178-043-1 52. Mohri, T., Toshiaki, T., Hozumi, Y., Hasegawa, T., Mochida, S., Tauchi, M., and Maruyama, S. (2001). Development on advanced high-temperature air combustion technology for steam reforming. Proceedings of Joint International Combustion Symposium, Hawaii, USA, September 9-12, 2001. 53. Suzukawa Y, Sugiyama S., Hino Y., Ishioka M., Mori I., Energy Convers. Mgmt. 38(10-13) (1997) 1061-1071. 54. Quinqueneau A., Touzet A., Oger M., Experimental studies on regenerative industrial burners operating in the flameless oxidation mode, 13th IFRF Members Conference, Ijmuiden, The Netherlands, May, 2001. 55. Quinqueneau A., Miquel P. F., Wills B., Experimental studies in an industrial staged burner with preheated air, 12 th IFRF Members Conference, Nooordwijkerhout, The Netherlands 5-8 May, 1998. 56. A. L.Verlaan, S. Orsino, N. Lallemant and R. Weber. Fluid flow and mixing in a furnace equipped with the low-NOx regenerative Burner of Nippon Furnace Kogyo. The results of the HAC97 trials. IRFR Doc. No.F46/y/1,1998. 57. R. Weber, Orsina S., Verlaan A.L. and Lallemant N., Combustion of light and Heavy Fuel Oils in high temperature air. Journal of the institute of Energy, June 2001, in printing. 58. S. Orsino, M. Tamura, P. Stabat, S. Costantini, O. Prado. and Weber R., Excess enthalpy combustion of coal, The results of the HTAC99 trials IFRF Doc, No.F46/y/3,2000. 59. R Weber, Combustion of natural gas, oil and coal with air preheated to temperatures in excess of 1000oC, 13th IFRF member’s conference. 60. R. Weber, A.L. Verlaan, S. Orsino, N. Lallemant, J. Inst. Energy, 72 (1999) 7783. 61. K.C. Szabo, J.D. Massingham, Low NOx regenerative burners for reheating furnace applications, 155 62. S. Lille, W. Blasiak, M. Mörtberg, T. Dobski, W.H. Yang, Heat flux evaluation in a test furnace equipped with high temperature air combustion (HiTAC) technique, 2002 International Joint Power Generation Conference, Phoenix, A.Z., USA, June 24-26, 2002. 63. R Nabil W. Blasiak, Measurements of flue gases compositions and flame volume in high temperature air combustion furnace, Finnish—Swedish flame days 2002, September 24-25, Vaasa, Finland. 64. M. Mörtberg., R Nabil W. Blasiak, Measurements of temperature, heat flux and flue gas composition in HTAC flame, Challenges in Reheating Furnaces, 28-29 October 2002, London U.K.. 65. T. Ishii, C. Zhang, S. Sugiyama, Journal of Energy Resources Technology, 120 (1998), 276-284. 66. Y. Hino, C. Zhang, T. Ishii, S. Sugiyama (1998) Comparison of measurements and predictions of flame structure and NOx emissions in a gas-fired furnace. AIAA/ASME Joint Thermophysics and Heat Transfer Conference. 67. P.J. Coelho, N. Peters, Combustion and Flame, 124 (2001) 503-518. 68. B. Pesenti, P. Evrard, P. Lybaert, 2001, NOx production and radiative heat transfer from an autoregenerative flameless oxidation burner, 4th HTACG International Symposium, Rome, 26-30 November. 69. S. Orsino, R. Weber, Numerical simulation of combustion of natural gas with high-temperature air. Combustion Science and Technology. 2002, vol.168, pp134. 70. M. Mancini, R. Weber and U. Bollettini, Predicting NOx emissions of a burner operated in flameless oxidation mode, Proceedings of the combustion Institute, 29, 2002, in printing. 71. Yang W., W. Blasiak, Combustion and heat flux evaluation in a furnace equipped with a High-cycle Regenerative System, Challenges in Reheating Furnaces, 28-29 October, IOM Communications, London, UK. 72. Rafidi Nabil, Thermodynamic aspects and heat transfer characteristics of HiTAC furnace with regenerators, Doctoral thesis, Royal Institute of Technology, Sweden, 2005. ISBN 91-7178-223-0. 73. Delichatsios, M. A., Transition from momentum to buoyancy-controlled turbulent jet diffusion flames and flame height relationships, Combustion and Flame, 92 (1993) 349-364. 156 74. T.R. Blake, M. McDonald, An examination of flame length data from vertical turbulent diffusion flames, Combustion and Flame, 94 (1993) 426-432. 75. H.A. Becker, P. Liang, Visible length of vertical free turbulent diffusion flame, Combustion and Flame, 32(1978) 115-137. 76. Ricou; F. P.; Spalding, D. B. J. Fluid Mech., 1961, 11, 21-32. 77. Hill, B. J.; J. Fluid Mech., 1972, 51, 773-779. 78. Becker, H. A.; Yamazaki, S. Entrainment, momentum flux and temperature in vertical free turbulent diffusion flames, Combustion and Flame, 1978, 33, 123149. 79. Zukoski, E. E.; Kubota, T.; Cetegen, B. Entrainment in fire plumes, Fire Safety Journal, 1980, 3, 107-121. 80. Cetegen, B. M.; Zukoski, E. E.; Kubota, T. Entrainment in the near field of fire plumes, Combustion Science and Technology, 1985, 39, 305-331. 81. Delichatsios, M. A.; Orloff, L. Entrainment measurements in turbulent buoyant jet flames and implications for modelling, Twentieth symposium (International) on Combustion, The Combustion Institute, Pittsburgh, 1984, p.367-375. 82. Han, D.; Mungal, M. G. Direct measurement of entrainment in reacting/nonreacting turbulent jets, Combustion and Flame, 2001, 124, 370-386. 83. Delichatsios, M. A. Air entrainment into buoyant jet flames and pool fires, Combustion and Flame. 1987, 70, 33-46. 84. Blake, T. R.; Cote, J. B. Mass entrainment, momentum flux, and length of buoyant turbulent gas diffusion flame, combustion and Flame, 1999, 117, 589599. 85. Muniz, L.; Mugnal, M. G. Effects of heat release and buoyancy on flow structure and entrainments in turbulent nonpremixed flames, Combustion and Flame, 2001, 126, 1402-1420. 86. Chigier, N. A.; Strokin, V. Mixing processes in a fire turbulent diffusion flame, Combust. Sci. Technol. 1974, 9, 111-118. 87. Takagi, T.; Shin, H. D.; Ishio, A. Properties of turbulence in turbulent diffusion flames Combust. Flame 1981, 40, 121-140. 88. Westerweel, D.; Dabiri; Gharib, M. The effect of a discrete window offset on the accuracy of cross-correlation analysis of digital PIV recordings, Exp. Fluids, 1997, 23, 20–28. 157 89. Clemens, N. T.; Paul, P. H. Effects of heat release on the near field flow structure of hydrogen jet diffusion flames, Combustion and Flame, 1995, 102,271-284. 90. Han D., Mungal, Simultaneous measurements of velocity and CH distribution. Part 1: jet flames in co-flow, Combustion and Flame, 132 (2003) 565-590. 91. T. Ishii, C. Zhang, S. Sugiyama, 2000, Journal of Energy Resources Technology, 122 (2000) 224. 92. W. Yang, W. Blasiak, Scandinavian Journal of Metallurgy, 33 (2004) 113. 93. P.C. Malte, D.T Pratt, Proceedings of the 15th symposium(international) on Combustion, , the combustion Institute , p1061-1070, Tokyo, Japan, August 2531, 1974. 94. G. Löffler, V. J. Wargadalam, F. Winter, H. Hofbauer, Combsution and Flame, 120 (2000) 427. 95. Z.E. Coda, P. Kilpinen, Energy & Fuels, 15 (6) (2001) 1425. 96. R. C. Steele, P.C. Malte, D.G., Nicol, J.C., Kramlich, Combust. Flame, 100 (1995) 440. 97. R. S. Barlow, G. J. Fiechtner, C. D. Carter, J.-Y Chen, Combust. Flame, 120 (2000) 549. 98. S. M. Correa, Combust. Sci. and Tech., 87 (1992) 329. 99. D. Tobacco, C. Innrella, C. Bruno, Combust. Sci. and Tech, 174(7) (2002) 1 100. Lille, S.(2002). Experimental study of single fuel jet combustion and high-cycle regenerative system. Doctoral thesis, Royal institute of technology, Stockholm.ISBN91-7283-300-9. 101. Magnussen, B.F.and Hjertager, B.H., (1977), On mathematical models of turbulent combustion with special emphasis on the soot formation and combustion. Proc. 16th Symp. (Int.) on Combustion, the Combustion Institute, pp.719. 102. Westbrook, C. K. and Dryer, F. L., (1981), Simplified reaction mechanisms for the oxidation of hydrocarbon fuels in flames. Combustion Science and Technology, vol. 27, pp.31-43. 103. D.L. Baulch, D.D. Drysdall, D.G., Horne, Evaluated kinetic data for High Temperature Reactions, Butterworth, 1973 104. Malte P.C. and Pratt D. T., (1974), Measurement of atomic oxygen and nitrogen oxides in jet-stirred combustion. Proceedings of the 15th symposium(international) on Combustion, pp1061-1070 158 105. G.G. De Soete, Proceedings of the 15th Symp. (Int.) on Combustion, The Combustion Institute, p.1093-1102. 1975. 106. Missaghi, M., Pourkashanian, M., Williams A. and Yap, L., 1990, The prediction of NO emissions from an industrial Burner, Proceeding of American Flame Days Conference, San Francisco. 107. G.G. De Soete, Proceedings of the 15th Symp. (Int.) on Combustion, The Combustion Institute, p.1093-1102. 1975 108. Löffler G. and Wargadalam V. J., Winter F. and Hofbauer H., (2000). Decomposition of nitrous oxide at medium temperatures. Combustion and Flame, Vol. 120, pp427-438 109. J.C. Hewson, M. Bollig, http://www- mae.ucsd.edu/cecr/Mechanisms/skeletal_table.html, CECR Research. 110. M. Frenklach, T. Bowman, G. Smith, B. Gardiner, http://www.me.berkeley.edu/gri_mech/data/frames.htm, GRI Mechanism 3.0. 111. Tabacco Daniele, Cuoco Francesco and Bruno Claudio, (2001). Theoretical and numerical investigation on flameless combustion and its potential for applications to gas turbines. Proceedings of 4th HTACG 2001 High temperature air combustion and gasification, Nov. 28-30, 2001,Room, Italy. 112. C.T. Bowman, Progress in Energy and Combustion Science 1 (1975) 33. 113. Star CD manual, Methodology Version 3.15, Computation Dynamics Limited 2000. 114. .Irvin Glassman, Combustion. Academic Press Inc., 1996 115. Hasselbrink E.F., Mungal M.G., Observations on the stabilisation region of lifted non-premixed methane transverse jet flames, 27th Symposium (International) on Combustion, The Combustion Inst., Pittsburgh, PA, 1998, pp. 1167 – 1173. 116. Stephen R. Turns, An introduction to combustion; concepts and applications, ISBN 0-07-116910-5, McGraw-Hill Co, Singapore, 2000. 117. Hawthorne, W. R.; Weddell, D. S.; Hottel, H. C. Mixing and Combustion in Turbulent Gas Jets. In Third Symposium on Combustion and Flame, and Explosion Phenomena; Williams and Wilkins: Baltimore, MD, 1949; p 266. 118. Strahle WC. Progress in Energy Combustion Science, Vol. 4,pp.157-176. 1978. 159 119. Baukal CE and Eleazer P B, 1998, Quantifying NOx for industrial combustion processes, Journal of Air Waste Management Association, January 1998, 48,5258. 120. Staab, P, Lockwood F.,1999, Application of mathematical modelling of full scale Industrial process: Status and need for sub-Models, Summary of the 12th TOTEM-Meeting of the IFRF, Ijmuiden, January 1999, IFRF Doc No D121/y/1. 160