APPLICATION OF A COAL COMBUSTION MODEL IN THE DESIGN OF BLAST PARAMETERS FOR AN IRONMAKING BLAST FURNACE

Transcription

1 Fifth International Conference on CFD in the Process Industries CSIRO, Melbourne, Australia 3-5 December 006 APPLICATION OF A COAL COMBUSTION MODEL IN THE DESIGN OF BLAST PARAMETERS FOR AN IRONMAKING BLAST FURNACE D. MALDONADO, P. ZULLI, Y.S. SHEN, B.Y. GUO and A.B. YU BlueScoe Steel Research Laboratories, P.O. Box 0, Port Kembla, NSW 505, AUSTRALIA Centre for Simulation and Modelling of Particulate Systems, School of Materials Science and Engineering, University of New South Wales, Sydney, NSW 05, AUSTRALIA ABSTRACT With significant economic drivers to reduce consumtion of exensive coking coal, Pulverized Coal Injection (PCI) commenced at BlueScoe Steel in 00, at injection rates ranging between 00 and 50 kg-coal/tonne of liquid iron. The key limitation to injection rates is associated with the reduction in acked bed ermeability via additional char load into the furnace. The coal is injected via a simle coaxial lance, consisting of an inner ie (for coal and carrier gas) and an outer annulus (for cooling gas to rotect the lance from the high furnace temeratures). The cooling gas can be comressed air, natural gas or ure oxygen. Deending on the choice of cooling gas, the oxygen-to-carbon ratio of the system will change. In this aer, the alication of a validated three-dimensional numerical model of the blowie/tuyere/raceway is described. The model is used for various lant-secific investigations of blast arameters such as oxygen enrichment, blast temerature and atomic oxygen-tocarbon ratio. The model results show the sensitivity of coal burnout to different oerating arameters and confirm that burnouts higher than 80% are difficult to obtain due to the short residence times of the coal. NOMENCLATURE A Arrhenius rate constant B Coal burnout C D Gas-solid drag coefficient d Particle diameter e Char article void fraction k Boundary layer diffusion (Gibb equation) k Surface reaction rate (Gibb equation) k 3 Internal diffusion rate (Gibb equation) m a Ash mass fraction m a,0 Original ash mass fraction M C Molecular weight of carbon Mo Molecular weight of oxygen molecule Nu Nusselt number Pr Prandtl number Re Reynolds number T c Activation temerature (K) T Particle temerature (K) ε Particle emissivity φ Molar ratio (Gibb equation) ρ Density σ Stefan-Boltzmann constant INTRODUCTION Pulverised coal injection was introduced at BlueScoe Steel s Port Kembla blast furnaces (BF) in 00, with aim injection rates in the range of kg-coal/tonne of liquid iron. Due to recent changes in suly coal ricing structure, there is a clear economic advantage associated with increasing PCI rates to relace coking coal. Overseas exerience, eg from Euroean steelmakers such as Corus IJmuiden and Arcelor Sidmar, suggest that coal injection rates u to and in excess of 00 kg-coal/tonne of liquid iron are sustainable. Combustion of coal under blast furnace conditions differs from the combustion henomena exerienced in coal fired ower stations because of the differences in air temerature, air flow rate, oxygen enrichment and nozzle geometry. Modelling of coal combustion under blast furnace conditions is relatively well advanced (Ishii, 000). One of the key factors affecting the maximum PCI rate attainable at the BF is the acked bed ermeability a more ermeable bed is more able to coe with the additional char load. In order to minimise char load, the extent of coal combustion (burnout) within the inlet nozzle (tuyere) and raceway must be otimised. This can be achieved through imrovements in a) lance design, eg double eccentric lances and/or lance ti shae, b) injection air (blast) arameters, eg oxygen enrichment, cooling gas tye, and c) coal tye. Recently, BlueScoe Steel Research and University of NSW jointly develoed a three-dimensional numerical model to describe the flow and combustion of ulverised coal under BF conditions (Guo et al., 005). The model was based on the geometry of, and flow conditions used in, BHP Billiton s third generation combustion test rig, located at its Newcastle Technology Centre (Mathieson et al., 003). The original version of the model was develoed using the commercial CFD ackage CFX ver4.4. More recently, a new version of CFX, ver0 (ANSYS CFX, 006), has been released with an imroved coal combustion sub-model within its solver architecture. In addition, the revious version of the model was relatively insensitive to atomic oxygen to carbon (O/C) ratios. This has been addressed by introducing the artial oxidation of char (carbon) to CO, as well as the gasification reactions of char with CO and H O. Homogeneous gas hase oxidation reactions of CO and H are also introduced.

2 MODEL DESCRIPTION Governing Equations for Gas Phase The gas hase flow field is described by transort equations of the continuous hase, i.e., three-dimensional, steady-state Reynolds-averaged Navier-Stokes equations, closed by the RNG k-ε turbulence model equations, based on the framework of software ackage ANSYS-CFX0. Governing equations solved for the gas hase include mass, momentum, turbulence kinetic energy, turbulence dissiation rate, enthaly, and a number of secies mass fractions. Particle Transort Pulverised coal articles are treated as a disersed hase using the Lagrangian method, by which article behaviours are calculated along the discrete article trajectories without interaction between articles. The trajectories of the discrete articles are determined by integrating Newton s second law, where the drag force is the main force comonent. The drag factor C D is given by Schiller and Naumann (933) as a function of Reynolds number Re: C = 4( Re ) / Re () D A number of reresentative articles are tracked sequentially, one-by-one, in a steady state simulation. The coal articles are treated as non-interacting sheres, each with uniform temerature and comosition. Mass conservation is maintained for each article; however, there is no allowance made for a change in article volume due to swelling and chemical reaction, ie the internal orosity of the article increases with mass loss. Full couling of mass, momentum and energy of articles with the gaseous hase is carried out. Particle disersion is included in the simulation. Devolatilisation Coal combustion consists of a multi-stage rocess: the devolatilization of a raw coal article, followed by the gaseous combustion of volatile matter (VM) and the oxidation of the residual char with the oxidant in the gas hase. Devolatilisation is tyically modelled as a two-ste rocess (Ubhayakar et al., 976), in which two reactions with different rate arameters and volatile yields comete to yrolyse the raw coal, as shown below. Coal k α V + α) ( Ch k α V + ( α ) Ch The first reaction dominates at lower temeratures with a lower yield α than the yield α of the second reaction (dominating at temeratures >~00 K). The yield of volatiles at higher temeratures, α, deends strongly on temerature and heating rate. A coal deendent Q-factor, or enhancement factor of volatiles yield, is used to adjust the volatile yield, while the low temerature devolatilisation yield, α, is based on the roximate VM content (dry ash free basis). Tyically Q-factors vary in the range of -.6. The rate constants k and k are article temerature (T ) deendent and have an Arrhenius form: k = Aex( T c T ) (3) () The default CFX ver 0.0 Arrhenius rate constants and activation temeratures, were used. These are summarised below. Reaction A [s - ] T c [K] Low rate (k ) High rate (k ) Table : Coal devolatilisation reaction kinetics Volatile Matter Comosition The VM consists of C, H and O comonents, i.e. it is assumed that it has a comosition of C x H y O, where x and y are a function of ultimate and roximate analyses of the resective coals, as well as a function of the enhancement or Q-factor. In the current study, an interim ste is included which assumes the comosition of C x H y O to consist of H, CO and a hydrocarbon, such that: C x H y O C x- H y-z + z H + CO (4) The different yields of hydrogen or hydrocarbon roduct are deendant on the ultimate and roximate analyses of the coal. Gas Phase Combustion The three volatile matter comonents were assumed to combust with oxygen, as follows: C α H β + (α+β/4) O α CO + β/ H O (5) H + ½ O H O (6) CO + ½ O CO (7) In addition to the above gaseous secies, nitrogen, N, is also resent in the system. The gas comositions are obtained from the solution of transort equations for the mass conservation of each of the comonents. Char Oxidation The char combustion model roosed by Gibb (985) was used in the current modelling work. Gibb s model takes into account the diffusion of oxygen within the ores of the char article as well as boundary layer diffusion and reaction kinetics. In this study two char combustion reactions are considered. They consist of: C + ½ O CO (8) C + O CO (9) In addition, CO formed during the artial oxidation of char (Eqn. 8) is assumed to combust to CO (Eqn. 7). The inclusion of the two char oxidation reactions increase the sensitivity of the results to oxygen. In the Gibb model, the oxidation mechanism of carbon can be characterised by the molar ratio φ of carbon atom/oxygen molecules, so that oxides are roduced according to the equation: φc + O CO (0) ( φ ) CO + ( φ)

3 From the above equation it can be seen that as φ, the main roduct consists of CO (Eqn. 9) and as φ, the main reaction roduct consists of CO (Eqn. 8). The molar ratio φ is resent in the reaction rate develoed by Gibb to calculate the rate of decrease in char mass m c : dm dt C φ ρ = k () 3 M C O, ( k + ( k + 3) ) e M O ρc From Eqns. 0-, it can be seen that the reaction which generates CO as the main roduct will have a reaction rate which is double the rate of the reaction which generates CO due to the resence of the molar ratio term φ in the numerator. Char Gasification Two additional heterogeneous gas-char reactions are considered in the current model formulation. These reactions consist of char gasification with CO (solution loss reaction) and H O (water gas reaction) as follows: C + CO CO () C + H O CO + (3) H At the high temeratures resent in the combustion of coal with enriched oxygen, the above reactions are exected to be limited by external and/or internal diffusion and not by reaction kinetics. As a result, a similar model as the Gibb combustion reaction was adoted to reresent the CO and H O char gasification reactions. The reaction rate was thus calculated with an equation of the same form as Eqn.. The reaction kinetics were based on char gasification studies conducted by CSIRO at 00 o C (Hla et al., 005). The CSIRO gasification studies comared the reactivity of different chars to CO and H O; the chars were derived from different Australian coals in a ressurised entrained flow reactor. The equation used in the model to calculate the char oxidation rate was of a modified Arrhenius equation of the form: k = A T ex( T T) (4) c c The values for the Arrhenius arameters are summarised in Table. Reaction A c [m s - K - ] T c [K] C- CO gasification C-H O gasification Table : Char gasification reaction kinetics Heat Transfer Between Particle and Gas Phase Both radiation and convection heat transfer are included in the model. The net radiative ower absorbed by a article is calculated by: 4 q = ε πd ( I σt ) (5) r The value of the article emissivity ε is assumed to vary linearly from the raw coal value to the char value as yrolysis roceeds. Convective heat transfer due to temerature difference between the fluid and articles is calculated using the Ranz-Marshall correlation (95): Nu = Re Pr (6) SIMULATION CONDITIONS Coal Proerties Five different ulverised coals with wide ranging roerties have been used in this study, with volatile matter contents ranging from 5.8 to 39.% (air dried, ad). The roximate and ultimate analyses of the coals are given in Tables 3-4, resectively. It is worthwhile noting that as VM increases, the oxygen content (lus errors) of the coal also increases. This increase is accomanied by a decrease in carbon content, which exlains the reduction in gross secific energy. Coal Coal Coal 3 Coal 4 Coal 5 Moisture, % (ad) Volatile matter, %(ad) Ash, % (ad) Fixed carbon, % (ad) Gross secific energy, 33,350 3,975 3,050 3,50 30,470 kj/kg (ad) Q-factor Table 3: Proximate analysis of coals Coal Coal Coal 3 Coal 4 Coal 5 C H N S O (by diff) Table 4: Ultimate analysis of coals (%) The mean article sizes for the different coal samles varied between 9-54μm. Laser diffraction analyses are summarised in Fig.. Cumulative volume % 00% 90% 80% 70% 60% 50% 40% 30% 0% 0% 0% Particle size, μm Coal Coal Coal 3 Coal 4 Coal 5 Figure : Laser diffraction sizing results for different coals 3

4 In the simulations, for each of the coals, 47 article size classes were samled in the range of to 50 μm, and a total of 47 reresentative articles were tracked. Geometry and Boundary Conditions Two different geometries were considered: the BHP- Billiton coal combustion test rig and one of Port Kembla s blast furnace tuyeres. Details of the test rig used to generate the exerimental data are described elsewhere (Guo et al., 005; Mathieson et al., 003). The main dimensions of the rig and the comutational mesh are shown in Fig.. This is the same mesh used in the study of Guo et al. (005). Figure : Main dimensions (mm) of the test rig model and outline of comutational mesh The comutational model was validated using the above geometry by comaring the redicted coal burnouts with measured values for a variety of coal tyes tested (Rogers et al., 006). Once validated under the test rig conditions, the model was alied to a geometry consisting of a blast furnace blowie/tuyere tyical of the Port Kembla furnaces, with the aroriate furnace oerating arameters used as boundary conditions. This geometry is shown in Fig. 3. a) Total blast flow, Nm 3 /h Blast O, % Coal flow rate, kg/h Cooling gas flow Coal Air, 3. Nm 3 /h Coal Air, 3. Nm 3 /h Coal Air, 3. Nm 3 /h Coal Air, 3. Nm 3 /h Coal Air, 3. Nm 3 /h Coal Methane,.8 Nm 3 /h Table 5: Summary of boundary conditions for cases considered in the test rig configuration The cases for the raceway geometry under blast furnace flow conditions considered the effect of: Coal tyes; Cooling gas tye (comressed air/oxygen/methane); Atomic oxygen-to-carbon ratio; Blast temerature; and Particle size distribution. RESULTS Coal Model Validation The validation of the coal combustion model formulation was based on a comarison between calculated and measured coal burnouts, with the latter obtained from exerimental data after Rogers et al. (006). The geometry of the test rig is shown in Fig.. Coal burnout, B, reresenting the total weight loss of the organic fraction of the coal due to volatile release and char gasification, is calculated using an ash balance: ma,0 B = ( ) /( ma, 0 ) (7) m a The measured vs calculated burnouts are summarised in Table 6 for the resective coals. b) Figure 3: Main dimensions of the BF raceway model For both geometries, there are three different gas streams entering the domain, namely, conveying gas (nitrogen), cooling gas (either comressed air, oxygen or methane) and hot blast. The flowrates for all three streams are based on exerimental and BF oerating conditions. For the test rig simulations, the cases considered were based on the resective coal tyes for air-cooling. An additional run was undertaken with methane as the coolant for Coal 4, the base case coal (3.5% VM). The boundary conditions for the six cases undertaken for the test rig geometry are shown in Table 5. Inlet blast temerature for all cases was 00 o C. BB (measured) (%) B (calculated) (%) Coal Coal Coal Coal Coal Coal 4 (methane cooling) Table 6: Comarison between exerimental and calculated burnout With the excetion of Coal (low VM, anthracitic coal), the comarison shows excellent agreement and indicates that the model assumtions and formulation are, within exerimental error, very good. The anthracitic coal was a articularly oor burning coal, and ossible exlanations for its oor erformance include low char reactivity, high roortion of coarser articles and oor devolatilisation characteristics. 4

5 Blast Furnace Raceway Geometry and Conditions Based on the same combustion arameters used in the validation of the comutational model, simulations were comleted for the raceway geometry (Fig. 3). Base case boundary conditions consisted of 300,000 Nm 3 /h blast flow, 35 t/h coal rate flow (Coal 4), 6000 Nm 3 /h oxygen enrichment in blast, 5000 Nm 3 /h oxygen cooling flow, 00 o C blast temerature and 45μm mean article size. The redicted flow field distributions of various arameters at the symmetry lane for the base case simulation are shown in Figs The article trajectories show very little disersion of the coal (Fig. 4). This is consistent with the high blast flow rates, the reduction in cross flow area at the tuyere nose and the adoted shae of the raceway. The coal ash mass fraction is highest near the uer erihery of the lume, which is consistent with the resence of smaller articles in this region and access to oxygen from the gas hase. Since the coal lume does not diserse significantly, the oxygen mass fraction distribution (Fig. 5) shows that the centre of the lume is oxygen-starved and the majority of the char combustion thus takes lace around the erihery of the lume. This is where the CO concentration and temerature are highest (Figs. 6-7). The distribution of CO shows a reduction in concentration near the end of the raceway. This is where CO and char are in close contact, allowing for the char gasification reactions to take lace. Figure 4: Particle trajectories Effect of coal volatile matter The simulation results for the five different coal tyes redict a high deendence of coal burnout with volatile matter content (Fig. 8). This is because of the much faster reaction kinetics associated with homogeneous hase combustion comared with the char oxidation reactions. Raceway geometry and BF conditions Burnout, % Figure 5: Oxygen mass fraction distribution Coal volatile matter, % Figure 8: Effect of coal tye on burnout for the BF geometry Effect of cooling gas tye The use of three different tyes of cooling gas was investigated: natural gas (3000 Nm 3 /h), comressed air (5000 Nm 3 /h) and oxygen (5000 Nm 3 /h). The main blast oxygen enrichment was maintained constant at 6000 Nm 3 /h. The burnout results are shown in Fig. 9. Figure 6: CO mass fraction distribution Figure 7: Temerature distribution Effect of cooling gas (raceway geometry) Burnout, % 7 Oxygen cooling 70 Comressed 68 air 66 Natural gas O/C atomic ratio Figure 9: Effect of cooling gas tye on coal burnout 5

6 A direct relationshi exists between the tye of cooling gas (denoted by atomic O/C ratio) and coal burnout. This is because in the case of natural gas cooling, the oxygen resent in the blast will combust with the natural gas as well as the VM released from the coal. As a result, the amount of oxygen resent to carry out the char combustion reaction will be reduced. Effect of blast temerature The effect of blast temerature was analysed, viz. 50 C, 00 C and 50 C. The redicted burnout results are shown in Fig. 0. Burnout, % Effect of blast temerature (raceway geometry) CONCLUSIONS A numerical model has been develoed and validated for the investigation of coal combustion henomena under blast furnace oerating conditions. It was found that: Consistent with exerimental data, the high Volatile Matter (VM) coals achieve much higher burnouts under simulated raceway conditions; A 7% (absolute) imrovement in coal burnout is redicted for a lance cooled by comressed oxygen, comared with a lance that is cooled with natural gas; Increasing the blast temerature by 50 C to 50 C resulted in a 4% (absolute) increase in coal burnout; The combination of very high oxygen enrichment and high coal flowrates results in high coal burnouts; and A reduction in the mean coal article size from 45μm to 9μm is redicted to increase burnout by 8% (absolute). ACKNOWLEDGEMENTS The authors would like to thank Dr Harold Rogers and his team at BHP-Billiton s Newcastle Technology Centre for the carefully carried out combustion exeriments and quality data rovided Blast temerature, o C Figure 0: Effect of blast temerature on coal burnout A direct relationshi exists between the blast temerature and the coal burnout. With the higher blast temeratures, coal devolatilisation takes lace closer to the lance ti. In addition the eak concentration of volatiles is higher. The resulting volatiles combustion generates additional enthaly to drive the downstream reactions and in addition, because of the earlier devolatilisation, there is a longer reaction time available in the raceway to consume the char in the slower heterogeneous reactions. Effect of additional variables Additional variables which were found to have an effect on coal burnout include the following. Blast oxygen enrichment The combination of very high oxygen enrichment and high coal flowrates results in high coal burnouts. Oerationally, this is similar to some Euroean blast furnaces. The high coal burnouts are a result of the high oxygen concentration available for volatile matter combustion, the generation of additional enthaly from the volatile matter combustion and a reduction in blast flow rates, which allows an increase in char residence time in the raceway. The reduction in blast flow rate is associated with the lower nitrogen inut into the furnace. Coal flow rate Burnout was directly roortional to coal flow rate. Particle size distribution The effect of coal article size distribution is also imortant. An imrovement in coal burnout of 8% (absolute) was redicted when the mean article size was reduced from 45μm to 9μm as a result of the enhanced coal devolatilisation and char reactivity. REFERENCES ISHII, K., (000), Advanced Pulverised Coal Injection Technology and Blast Furnace Oeration, Elsevier Science Ltd. GUO, B., ZULLI, P., ROGERS, H., MATHIESON, J. G. and YU, A., (005), Three-dimensional Simulation of Flow and Combustion for Pulverised Coal Injection, ISIJ International, 45, 7-8. MATHIESON, J. G., TRUELOVE, J. S. and ROGERS, H., (003), Toward an Understanding of Coal Combustion in Blast Furnace Tuyere Injection, Int. Sym. on Utilisation of Coal and Biomass, ISUCB-03 Newcastle, Australia. ANSYS CFX, CFX0.0, (006), on line document ( SCHILLER, L. and NAUMANN, A., (933), Uber die Grundlegenden Berechnungen bei der Schwerkraftaufbereitung, VDI Zeits, 77, 38. UBHAYAKAR, S. J., STICKLER, D. B., von ROSENBERG Jr., C. W. and GANNON, R. E., (976), Raid Devolatilization of Pulverised Coal in Hot Combustion Gases, 6th Sym. (Int.) Combust./The Combustion Institute, 47. GIBB, J., (985), Combustion of Residual Char Remaining after Devolatilization, Lecture, Mechanical Engineering Det, Imerial College, London. HLA, S., HARRIS, D. and ROBERTS, D., (005), A Coal Conversion Model for Interretation and Alication of Gasification Reactivity Data, Proc. Int. Conf. of Coal Sci. and Tech. (ICCS&T), Ibaraki, Jaan. RANZ, W. and MARSHALL, W., (95), Evaoration from Dros: Part I, Chemical Engineering Progress, 48, 4. ROGERS, H. (006), BlueScoe Steel Phase II Blast Furnace PCI Combustion Test Program, Internal BlueScoe Steel Reort. 6