Study of syngas co-firing and reburning in a coal fired boiler

Transcription

1 Fuel 83 (2004) Study of syngas co-firing and reburning in a coal fired boiler K.-T. Wu a, *, H.T. Lee a, C.I. Juch a, H.P. Wan a, H.S. Shim b, B.R. Adams b, S.L. Chen c a Energy and Resources Laboratories, Industrial Technology Research Institute, Bldg 64, 195, Chung-Hsin Rd, Sec 4, Chutung Hsinchu 310, Taiwan, ROC b Reaction Engineering International, 77 West 200 South, Suite 210, Salt Lake City, Utah 84101, USA c Pacific Rim Technologies, 50 Ln. 46 Ave 688, Jong Jeng Rd., Pingtung 900, Taiwan, ROC Received 31 December 2003; revised 15 March 2004; accepted 28 March 2004; available online on 27 April 2004 Abstract Combustion simulations were conducted to evaluate the technical feasibility of using a waste-based syngas as a supplemental fuel in a 70,000 kg/h-steam coal-fired boiler. The syngas was either co-fired with coal at the burners or injected downstream as a reburning fuel under both fuel lean and conventional (fuel rich) reburning configurations. Sensitivity of syngas heat input and furnace stoichiometry were examined. Results indicated that the syngas was an effective reburning fuel although it contained less than 6% hydrocarbons. NO x reductions from 12 46% were predicted for different reburning configurations; the highest NO x reduction of 46% was achieved with 23% heat input under a conventional reburning configuration. Furnace LOI increased over baseline values for all but one reburning configuration. The highest LOI came from the top burner row in all cases. Results suggested the LOI increase due to reburning might be mitigated by biasing more air to the upper burners to enhance particle burnout. Predictions indicated co-firing syngas at the burner centerline was not beneficial as it resulted in poor combustion of the coal particles and high LOI and CO emissions. Simulation results also indicated that minor changes to the syngas composition do not significantly affect the performance of syngas reburning. q 2004 Elsevier Ltd. All rights reserved. Keywords: Syngas; Co-firing; Reburning; NO x ; Combustion simulation 1. Introduction The co-firing of natural gas has generated significant interest among coal burning power producers in the United States [1]. Several full-scale studies have demonstrated both the technical and economic feasibility of co-firing natural gas as much as 20% by heat input. Because natural gas contains no ash and virtually no sulfur or nitrogen, co-firing natural gas in coal-fired boilers reduces SO 2, NO x and particulate emissions. However, natural gas is very expensive in Taiwan. In addition, gas pipelines are not readily available in most areas. The gasification of biomass to produce synthesis gas (syngas) on-site offers an option for fossil fuel fired boilers in Taiwan. The syngas can be used as a supplemental fuel to reduce the consumption of imported fuels such as pulverized coal and fuel oil. Furthermore, since it contains hydrocarbons and other reducing compounds such as * Corresponding author. Tel.: þ ; fax: þ address: ktwu@itri.org.tw (Keng-Tung Wu). hydrogen and carbon monoxide, it has the potential to be used as a reburning fuel to reduce NO x emissions [2]. The Energy and Resources Laboratories of Industrial Technology Research Institute in Taiwan has developed a circulating fluidized bed (CFB) gasifier to produce syngas out of paper rejects from paper mills [3]. The feasibility of using the syngas as a supplemental fuel is planned to be demonstrated in a coal-fired boiler in Taiwan. There are several uncertainties, however. The syngas produced from the paper rejects is of low heat content, ranging from 3.99 to 6.24 MJ/kg [4]. In addition, the syngas contains trace amounts of NH 3, HCN, HCl and H 2 S. Their impacts on boiler operations and emissions should be assessed. Test burns are expensive to conduct and run the risks of equipment damages and life shortening on the boiler. Computer simulation is a cost effective way to evaluate the impacts of fuel switching or equipment modification prior to field retrofits. This paper describes a feasibility study with computer simulations to evaluate syngas as a supplemental fuel for a coal-fired industrial boiler. The focus of the study is on the impacts of syngas co-firing and reburning on NO x and CO emissions and unburned carbon in fly ash /$ - see front matter q 2004 Elsevier Ltd. All rights reserved. doi: /j.fuel

2 1992 K.-T. Wu et al. / Fuel 83 (2004) Fig. 1. Schematic of the two boiler sections radiant furnace and backpass. 2. Boiler and syngas simulated The boiler selected for the study is a front wall-fired unit with 70,000 kg/h steam capacity. Fig. 1 shows the radiant furnace geometry of the unit. There are a total four burners. The burners on each row swirl in opposite directions. Each burner is equipped with a coal impeller on the centerline. Fig. 1 also shows that the convective section (backpass) consists of a series of superheaters and economizers. Coal properties used for the analyses are shown in Fig. 2. The particle size distribution was divided into discrete mass fraction bins to represent the distribution. The syngas composition is presented in Table 1 [4]. Two syngas compositions are shown. Case 1 through Case 4 used the initial composition. Case 5, a parametric case designed to evaluate the sensitivity of the results to syngas composition, used the second syngas composition. 3. Model description A computational fluid dynamics (CFD) software package GLACIER [5 8] from Reaction Engineering International (REI) was used for this study. These computational tools have been used extensively to model the physical and chemical processes occurring in industrial furnaces, in particular, utility and industrial fossil fuel-fired boilers [9 12]. Particular emphasis has been placed on simulating coal combustion and pollutant formation. The fluid dynamics of the gases and particulate matter flowing through a combustion chamber are computed to determine the convective and diffusive mixing of mass, momentum, energy and relevant chemical species. The flow and chemical reactions are influenced by the radiative heat transfer, which is generated by the combustion process. The radiative heat transfer influences the local devolatilization Fig. 2. Coal properties.

3 K.-T. Wu et al. / Fuel 83 (2004) Table 1 Syngas composition and heating value [4] Cases 1 4 Case 5 Moisture %, wet N 2 %, dry H 2 %, dry CO %, dry CO 2 %, dry CH 4 %, dry C 2 H 2 %, dry C 2 H 4 %, dry C 2 H 6 %, dry C3 C5 %, dry NH 3 ppmv, dry HCN ppmv, dry H 2 S ppmv, dry COS ppmv, dry 5,0.5 HCl ppmv, dry Tars þ benzene g/dnm and heterogeneous combustion of the particles, the local gas temperature and thus the local fluid dynamics. The influence of the deposition of ash residue on heat transfer surfaces and the effect on the radiation field is also included. Particle trajectories are followed by introducing them into the flow field which is induced by the mixing and reaction processes. Their position in turn influences the velocity, pressure, concentration and temperature fields of the combustion process. Computations include full mass, momentum and energy coupling between the gas and particles as well as full coupling between turbulent fluid flow, chemical reactions and radiative and convective heat transfer. GLACIER solves the governing gaseous fluid mechanical and reaction equations in an Eulerian framework. Turbulence is incorporated with a k-1 turbulence model for closure that has been modified for the presence of particles. The gaseous reaction rates are assumed to be mixing limited. Gas-phase mixture fractions are defined for the local mass fraction of inlet carrier gas fuel and the local mass fraction of gas evolved from the solid coal particles. Transport equations are solved for each mean mixture fraction and its variance about the mean. Statistical probability density functions (pdf s) are used at each point in the flow field to obtain mean properties of chemical composition, temperature, and other variables based on local instantaneous equilibrium and then by convolution of the pdf [5]. The particle mechanics are solved by following statistically described trajectories (means and variations) for a discretized group or cloud of particles in a Lagrangian frame of reference [7]. Particle reaction processes include coal devolatilization, char oxidation and gas-particle interchange. Particle swelling is accounted for empirically. The particles are assumed to be isothermal. Particle reaction rates are characterized by multiple parallel reaction rates with fixed activation energies. The parameters which describe the particle reaction rates are part of the input to Table 2 Coal devolatilization rate expressions and parameters Two step model Raw coal! k1 Y 1 (volatiles) þ (1 2 Y 1 ) (char)! k2 Y 2 (volatiles) þ (1 2 Y 2 ) (char) Rate form k i ¼ Ai exp ð2e i =RTÞ Reaction Y i A i (s 21 ) E i (kcal/mol) the code. In this case, coal devolatilization was modeled using a two-step model proposed by Ubhayakar [18]. The kinetic parameters used for this furnace are shown in Table 2. A global Arrhenius model was used to model heterogeneous char oxidation. The kinetic parameters used in this simulation are shown in Table 3 [19]. Particles are defined to consist of coal, char, ash and moisture. Ash is inert by definition; volatile mineral matter is considered as part of the volatile matter of the coal. The off-gas from particle reactions is assumed to be of constant elemental composition. Turbulent fluctuations and complete, local, complex chemical equilibrium are included in the particle reactions. Heat, mass, and momentum transport effects are included for each particle. The code models particle-laden flame radiation including effects of multicomponent, non-uniform, emitting, absorbing, scattering surface and medium properties. Mechanisms modeled include nonisotropic scattering from the particles, absorption and emission due to sooting, and spectrallybased CO 2 and H 2 O absorption and emission. The transport of thermal radiation is solved with a discrete ordinates model [8]. The solution algorithm for two-phase flow employs a series of macro- iterative loops over the particle and gas calculations. Within each gas phase macro-iteration loop, an iterative loop is performed over the governing partial differential equations (PDE) for fluid mechanics, chemistry (i.e. fuel stream mixture fractions and their variances), and radiative transport in a sequential manner to obtain updated solution values. The governing equations for the gas phase are solved using the SIMPLER method a pressure-based, segregated variable scheme developed for low speed, variable density flows [25]. Gas properties are updated by first computing the local mean mixture fraction and parameterized heat loss variables, and then computing the local mean thermo-chemical properties of the gas. Table 3 Char oxidation rate expression and parameters k ¼ A exp ð2e=rtþ A (gc/cm 2 s atm) E (kcal/mol)

4 1994 K.-T. Wu et al. / Fuel 83 (2004) The particle phase is computed by solving an initial value problem for the mean trajectory and dispersion of the cloud of particles. Included in the system of ordinary differential equations (ODE) are equations for: particle momentum, continuity of species, particle energy, particle liquid vaporization, particle devolatilization, and char oxidation. The governing set of ODEs is solved using a time-accurate predictor corrector method for stiff ODEs. Calculation convergence is based on the minimization of a series of error residuals for the governing PDEs. A calculation is considered to be converged when iteration to iteration residuals show that all PDEs are solved to within five digits of accuracy. Other factors such as overall mass balance, energy balance and extent of particle reactions are also used to measure convergence. Gas-phase underrelaxation factors of 0.65 were used with the SIMPLER solver in this study. The rates governing the formation and destruction of NO x are significantly slower than those governing the primary heat release reactions. Thus, the assumption that these species are in local chemical equilibrium leads to inaccurate predictions of their local concentrations. To account for the finite-rate of formation/destruction of these species, additional rate equations are solved. The NO x chemistry has a negligible effect on the local temperature and velocity fields; hence this analysis is de-coupled from the solution of the turbulent flow field and performed as a post-process for computational efficiency. Computing full finite-rate chemistry for all intermediates in a turbulent coal or gas fired furnace is beyond the capability of current combustion simulation tools. However, to obtain engineering estimates of the sensitivity of outlet NO x to design and operating variables in an industrial combustion system, this degree of detail within the model is not required. This objective can be met through the use of reduced descriptions of the chemical kinetics as long as all other physical mechanisms of first-order importance are also included in the analysis. The approach taken to develop reduced descriptions of NO x chemistry, is the use of conventional reduced mechanisms [20,21]. This approach is based on the assumption that certain species contained within a complete detailed chemical mechanism are in steady-state (i.e. their rate of production is equal to their rate of destruction). The total number of transfer equations that must be solved within the CFD simulation is equal to the number of major non-steady state species. This is a more rigorous and universal approach than curve-fitting type approaches which are often used to generate simple global mechanisms. The reduced mechanisms that are obtained are often much more robust, representing the detailed chemistry over a wider range of conditions than can be obtained with global mechanisms. The computer assisted reduction method (CARM) [22,23] automates the mechanism reduction process and produces FORTRAN source code for the calculation of the chemical source terms defined by the reduced mechanism. The inputs for the subroutine are pressure, temperature, and the mass fractions of the major species. The output is the reaction rate (or chemical source terms) for these major species. The detailed mechanism on which this reduced mechanism has been based was developed through an experimental and theoretical study of reburning and hybrid reburning/sncr chemistry in the temperature range K [24]. This mechanism consists of 438 reactions and 66 chemical species. The reduced mechanism retains 16 species (CH 4,CH 3,CO 2, CO, H 2,O 2, OH, H 2 O, C 2 H 2, C 2 H 6, N 2, NO, HCN, N 2 O, HNCO, and NH 3 ). This mechanism was found to produce excellent results, in comparison to the detailed mechanism, over the range of stoichiometric and temperature conditions of interest. 4. Modeling approach The boiler was divided into two parts as illustrated in Fig. 1 for modeling purposes. In the first part, a full CFD analysis of a radiant furnace was performed. This included evaluation of the combustion effects (flow patterns, flame shape, heat transfer) species profiles (e.g. CO, O 2,NO x, H 2 S, HCN, NH 3 ) in the radiant furnace and the unburned carbon, or loss on ignition (LOI) of the fly ash. In the second modeling part, a computational grid of the upper furnace was created including geometry for the superheat tubes, screen tubes and economizer tubes (see Fig. 1). Flow mixing, gas temperature profiles, and chemical reactions of the flue gas were tracked. The principal focus of this analysis was to estimate the extent of CO oxidation in the boiler backpass. This was done using finite-rate chemistry for CO reactions in the flue gas. The extent of CO oxidation is dependent on the initial amounts of CO entering the convective pass, the extent of mixing in the backpass and the temperature profile in the backpass. At temperatures greater than, 875 8C, CO oxidation is rapid and can be modeled with chemical equilibrium approaches. In the range from C, CO oxidation is slower and should be modeled with finite-rate kinetics. Below 700 8C, CO oxidation does not occur even with oxygen present. In order to evaluate the impacts of syngas, the following cased were simulated: Baseline (no syngas) Syngas co-firing (Case 1); Lean syngas reburning (Cases 2 and 3); Conventional syngas reburning (Case 4); and Syngas composition (Case 5). All of the cases were conducted with a half-furnace computational grid based on symmetry in the furnace (see Fig. 1). The half-furnace model had a total of 334,000 computational points. Similarly, the backpass modeling for

5 K.-T. Wu et al. / Fuel 83 (2004) each of the cases was conducted with a half-furnace (symmetric) backpass model. 5. Results and discussion The inputs and predicted results for the baseline and syngas co-firing cases are summarized in Table 4. The baseline case simulated the current boiler operation conditions. Fig. 3 shows the temperature and CO distributions in the furnace. Due to the high-swirl nature of the burner, the flames were quite wide and short, turning upward to the exit before impacting the rear wall of the furnace. This is consistent with observations in the field. The average CO concentration at the furnace exit was 3859 ppm (see Table 4). However, as shown in Fig. 4, the CO decayed very rapidly to less than 1 ppm in the backpass. This is because the flue gas was fuel lean; there were no fuel rich pockets to propagate the high CO levels through the backpass. Under these fuel lean conditions, the CO oxidation requires only time (or in this case distance) at sufficient temperature ( C) and normal turbulence, not bulk mixing of fuel lean and fuel rich pockets. These conditions are satisfied in the upward section of the backpass, and the CO is completely oxidized in this section. The measured CO concentrations in the stack were also very low (,10 ppm) although not zero. In the actual dynamic furnace, there may be some fuel rich pockets that occasionally enter the backpass. If this is the case, there could be a few ppm of CO that exit the backpass and reach the stack. The predicted NO x concentration was 460 ppm (dry, 6% O 2 ). This compares well with the measured value of 410 ppm in Table 4 Summary of model inputs and results Base Case 1 Case 2 Case 3 Case 4 Case 5 Stoichiometric ratio Exit O 2 %, dry Total firing rate, (10 6 kcal/h) Coal Feeding rate (kg/h) Firing rate (10 6 kcal/h) Firing rate (10 6 BTU/h) Syngas Heat replacement (%) NA Feeding rate (kg/h) NA Firing rate (10 6 kcal/h) Firing rate (10 6 BTU/h) Primary air Total air (%) Temp (8C) Flow rate (kg/s) Secondary air Total air (%) Temp (8C) Flow rate (kg/s) Overfire Air Total air (%) NA NA NA NA 26.9% NA Temp (8C) 260 Flow rate (kg/s) 6.05 Results Furnace exit temp (8C) Furnace exit CO, dry (ppm) , LOI (%) NO x, 6% dry (ppm) NO x, lb-no2/mbtu NO x (tons/yr) Burnout Furnace thermal efficiency (%) Furnace and backpass efficiency (%) Backpass exit temp (8C) Backpass exit CO, dry (ppm),1 78.4,1,1,1,1

6 1996 K.-T. Wu et al. / Fuel 83 (2004) Fig. 3. Baseline temperature and CO profiles in the furnace. the boiler. The predicted LOI of the fly ash was 6.5%, which is comparable with the measured 9.9% in the field. The baseline results serve as a basis for comparisons with the syngas co-firing cases to be discussed below Syngas co-firing A direct way to co-fire syngas in the boiler is to inject it through the pipe which holds the oil injector in the center of the burner. Due to the low heating value of the syngas, a significant mass flow would be required to achieve the desired heat input replacement. Originally 10% heat input replacement was desired, but the required mass flow was too large to put through the burner center pipe with a diameter of 90 mm, so the amount of syngas was reduced to 5% heat replacement for this case. This corresponded to an injection velocity of 88 m/s, which was still quite fast. Fig. 5 shows the effect of the injected syngas on the flame shape. Injecting the syngas in the burner core results in the elimination of the central recirculation zone just outside the burner and a much narrower flame. This flame does not mix as well as the baseline flame, resulting in lower NO x formation but also lower particle burnout (less fuel released in the furnace) and higher LOI and CO concentrations at the furnace exit (see Table 4). The average CO concentration at the furnace exit was much higher than the baseline (16,780 ppm vs ppm) and resulted in some CO emissions at the economizer exit (,80 ppm). The NO x reduction was due more to changing the flame behavior than any chemical reactions with the syngas. This is consistent with previous field demonstrations that have shown cofiring 10 20% natural gas in the burner zones of coal-fired boilers does not produce significant NO x reductions [13]. 6. Syngas reburning Reburning is another way to co-fire syngas with coal in the boiler. Reburning is a combustion modification technology which removes NO x from combustion products by using fuel as the reducing agent. Approximately 70 90% of the total heat input is provided by burning the main fuel (coal) under fuel-lean conditions. A reburning fuel, such as gas or coal, which provides the balance of the heat input, is injected downstream of the coal burners to create a reducing zone (reburning zone) which destroys the NO x produced from the combustion of coal. Since the reburning zone is under fuel-rich conditions (with conventional reburning), additional air is required to be injected downstream of the reburning zone to complete the combustion process. Over 50% NO x reductions have been achieved with natural gas and coal reburning in utility boilers [14]. Fuel lean reburning differs from conventional reburning in that all the air is added with the main fuel and thus Fig. 4. Average CO and temperature profiles over the mean flow path through the backpass.

7 K.-T. Wu et al. / Fuel 83 (2004) Fig. 5. Burner flow patterns for the baseline and in-burner syngas injection cases. Color represents oxygen concentration red is high concentration (,21%) and blue is low concentration (,0%). the available excess air limits the amount of reburning fuel. In practice, it is limited by CO emissions as the injected reburning fuel must mix with the available oxygen and react to complete combustion before CO oxidation ceases. Approximately 30% NO x reduction was achieved with fuel lean reburning when 6% natural gas was used as the reburning fuel in a (coal-fired) cyclone utility boiler [15]. Both conventional and fuel lean reburning were investigated with the syngas. Fig. 6 illustrates the firing configurations and the stoichiometry distributions for two fuel-lean reburning cases, Case 2 and Case 3, and one conventional reburning case, Case 4. For all cases, the reburning syngas was injected at 1.2 m above the top coal burners through two injectors on the front wall and four injectors on the rear wall. The heat input of the syngas was 10% for the two lean reburning cases and 23% for the conventional reburning case, respectively. For the conventional reburning case, over-fire air was injected at 1.7 m above the syngas through two injectors on both the front and rear walls to ensure adequate penetration and mixing with the combustion gases. Lean reburning with Case 2 resulted in only about 12% reduction in NO x emissions from the baseline (see Table 4). However, as the reburning zone stoichiometry was reduced from 1.30 to 1.15, approximately 30% reduction in NO x emissions was obtained with Case 3 under lean reburning conditions. As the reburning zone stoichiometry was further reduced to 0.95, the NO x emission was reduced by 46% with Case 4 under conventional reburning conditions. More NO x reduction can be expected if more syngas is added to reduce the reburning zone stoichiometry to , the optimum for conventional reburning [2]. In practice, this might be limited by the low heating value of the syngas and its availability, however. In order to understand the effectiveness of syngas reburning, the NO x concentrations along the furnace elevations were examined and presented in Fig. 7. Comparing the NO x levels before and after the syngas Fig. 6. Stoichiometry distributions for syngas reburning configurations (Cases 2, 3 and 4).

8 1998 K.-T. Wu et al. / Fuel 83 (2004) Fig. 7. Comparison of temperature, stoichiometric ratio and NO x values for the Case 2 (top), Case 3 (middle) and Case 4 (bottom) syngas reburning strategies. Values are based on area-weighted averages as a function of furnace elevation. Note different scales for different variables (a NO x value of 1800 on the plot is actually a NO x concentration of 1800/5 ¼ 360 ppm). injection for Case 2 (top plot) indicates that a relatively small amount of NO x reduction was achieved with syngas reburning. Similarly only slightly more NO x reduction can be attributed to syngas reburning alone for Case 3 (middle plot) although close to 30% reduction was observed in NO x emissions. It appears that significant amount of the reduction in these lean reburning cases is due to the lower NO x produced by the coal burners. On the other hand,

9 K.-T. Wu et al. / Fuel 83 (2004) Table 5 Summary of LOI in fly ash and contributions by burner row LOI Baseline Case 1 Case 2 Case 3 Case 4 Case 5 Top burner row (%) Bottom burner row (%) Overall (%) the data from Case 4 (bottom plot) indicates that the large majority of NO x reduction resulted from syngas reburning. Since the syngas contains less than 6% hydrocarbons (wet basis), some of the reductions might be attributed to the hydrogen [2] and CO [16] contents in the syngas (see Table 1). Although the syngas is not as effective as natural gas as a reburning fuel, it, unlike natural gas [14], will not require the use of a transport medium such as flue gas or steam to provide adequate penetration and mixing in boiler applications thanks to its high inert contents. The impacts of syngas reburning on carbon burnout are presented in Table 5. Case 2 had the lowest LOI (4.0%) of any case studied (including Case 5 which will be discussed in Section 6.1). This was due to the increased oxygen availability in the burner zone (as a result of the higher burner zone stoichiometry, 1.377). As the burner zone stoichiometry was decreased to 1.216, Case 4 produced much higher LOI (11.0%). It appears that the addition of syngas further reduced the oxygen availability to the char particles. Although Case 4 had the lowest burner zone stoichiometry, it did not produce the highest LOI because of the addition of the over-fire air. Table 5 also shows the contributions of each burner row to the overall LOI content of the fly ash. LOI results from unoxidized char in the coal particles. It is a function of several factors including residence time, oxygen, particle temperature, coal reactivity, and particle size. It can be seen from Table 5 that the majority of the LOI comes from the top burner row. In all cases, the top burner row has higher LOI, suggesting that some combination of shorter residence time and lower oxygen is restricting char oxidation in the particles coming from the top burners. This information is unique to CFD modeling (that is, it cannot be obtain via experimental methods) and is very helpful in determining a strategy for controlling LOI [17]. For example, it may be possible to bias air flow to the top burners in order to enhance char oxidation for the particles from this burner row. The bottom burners would remain at lower excess air levels, helping control NO x formation Impacts of syngas composition Since the CFB gasifier produces syngas with some variations in composition, Case 5 was conducted to study the impacts of syngas composition on reburning performance. Inputs and configuration for Case 5 were the same as Case 3, with only the syngas composition changed. As shown in Table 1, the most significant differences between the compositions are the decrease in HCl concentration and the increase in NH 3 concentration for Case 5. Case 5 syngas also has a 13% lower heating value, meaning more syngas must be fired to reach the same heat input. Reviewing the result summary presented in Table 4 indicates that Case 5 and Case 3 behaved almost identically. The high NH 3 concentration in the Case 5 syngas was a concern as it is a source of fuel NO. If all of the NH 3 in the Case 5 syngas were converted to NO, it could result in an increase of 15 ppm (dry, 6% O2) in NO x emissions in the flue gas. In addition, the Case 5 syngas contained slightly lower hydrocarbon, hydrogen and CO concentrations. All these potentially detrimental factors did not appear to lessen its reburning effectiveness, however. 7. Conclusions Predictions indicated co-firing syngas at the burner centerline was not beneficial as it resulted in poor combustion of the coal particles and high LOI and CO emissions. However, in spite of its low hydrocarbon content, the syngas was shown to be effective as a reburning fuel. NO x reductions from 12 46% were predicted for different reburning configurations; the highest NO x reduction of 46% was achieved with 23% heat input under a conventional reburning configuration. Furnace LOI increased over baseline values for all but one reburning configuration. The highest LOI came from the top burner row in all cases. Results suggested the LOI increase due to reburning might be mitigated by biasing more air to the upper burners to enhance particle burnout. Lean reburning with syngas can be considered for applications where significant NO x reductions are not required and there is not enough residence time in the boiler for the installation of overfire air ports. Simulation results also indicated that minor changes to the syngas composition do not significantly affect the performance of syngas reburning. Acknowledgements The authors acknowledge the financial support provided by the Energy Commission, MOEA, Taiwan, ROC.

10 2000 K.-T. Wu et al. / Fuel 83 (2004) References [1] Swanekamp R. Natural gas poised to penetrate deeper into electric generation. Power 1995; [2] Chen SL, McCarthy JM, Heap MP, Seeker, WR, Pershing DW. Bench and pilot scale process evaluation of reburning for in-furnace NO x reduction. 21st Symposium (Int.) on Combustion. Pittsburgh, PA: the Combustion Institute; p [3] Industrial Technology Research Institute. Development and dissemination of waste-to-energy technologies (FY2002). Final Report Prepared for Energy Commission, Taiwan, ROC; [4] Lee HT, Wu K-T, Juch CI, Tsai MY, Huang CC, Kylä-Sipilä M, Hiltunen M. Gasification of paper reject in a CFB pilot plant. In: Bridgwater AV, editor. Pyrolysis and gasification of biomass and waste. UK: CPL Press; p [5] Smoot LD, Smith PJ. Coal combustion and gasification. New York, NY: Plenum Press; [6] Smith PJ, Fletcher TH. A study of two chemical reaction models in turbulent coal combustion. Combust Sci Technol 1998;58:59. [7] Smith PJ. 3-D turbulent particle dispersion submodel development. DOE Quarterly Progress Report #4, US; [8] Adams BR, Smith PJ. Three-dimensional discrete-ordinates modeling of radiative transfer in a geometrically complex furnace. Combust Sci Technol 1993;88:293. [9] Eddings EG, Heap MP, Pershing DW, Sarofim AF, Smith PJ, Spinti JP. The design of a coal-fired high-temperature furnace for an advanced combined cycle system. Second International Conference on Combustion Technologies for a Clean Environment, Lisbon, Portugal; [10] Eddings EG, Davis KA, Heap M, Valentine J, Facchiano A, Hardman R, Grigas N. Effect of low NO x firing conditions on increased carbon in ash and water wall corrosion rates. Presented at the DOE/EPRI/ EPA Mega Symposium on NO x /SO x /Air Toxics, August [11] Bockelie MJ, Adams BR, Cremer MA, Davis KA, Eddings EG, Valentine JR, Smith PJ, Heap MP. Computational simulations of industrial furnaces. Computational Technologies For Fluid/Thermal/ Chemical Systems with Industrial Applications, San Diego, California; [12] Cremer M, Adams B, Wang D, Heap M. CFD modeling of NO x reduction technologies in utility boilers. FACT-Vol. 23/HTD-Vol. 367, Combustion, Fire, and Computational Modeling of Industrial Combustion Systems: ASME; p [13] SFA Pacific, Inc., California, Electric Power Technologies, Inc. Gas co-firing for coal fired utility boilers. ERPI Report, TR ; [14] Engelhardt D, Folsom B, Latham C, Moyeda D, Payne R, Sommer T, Brocato H, Maziarz M. Updated experience using reburn technology for utility boiler NO x emissions reductions. EPRI-DOE-EPA Combined Utility Air Pollution Symposium: The MEGA Symposium; [15] Adams BA, Wang D, Cremer M, Frizzell K, Conn S. Modeling NO x reduction from fuel lean gas reburning and selective non-catalytic reduction combined with overfire air at OMU s Smith Unit 1. EPRI/ DOE/EPA Mega Symposium, Chicago, Illinois; [16] Ho TY, Widmer NC, Chen SL. Study of COG and BFG reburning for the control of NO x emissions from coal fired boilers. Asia-Pacific Conference on Combustion, Osaka, Japan; [17] Chen JL, Li KC, Chen CF, Davis KA, Valentine JR, Heap MP, Chen SL. Simulation of burners out of service and coal co-firing in a utility boiler. Presented at Second Asia-Pacific Conference on Combustion, Tainan, Taiwan 1999; [18] Ubhayakar SK, Stickler DB, von Rosenberg CW, Gannon RE. Rapid devolatilization of pulverised coal in hot combustion gases. 16th Symposium (Int.) on Combustion 1977; [19] Baxter LL. Turbulent transport of particles. PhD Dissertation, Brigham Young University; [20] Smooke MD, Reduced kinetic mechanisms and asymptotic approximations for methane-air flames, Berlin: Springer; [21] Peters N, Rogg B, editors. Reduced kinetic mechanisms for applications in combustion systems, Berlin: Springer; [22] Chen JY. A general procedure for constructing reduced reaction mechanisms with given independent relations. Combust Sci Technol 1998;57: [23] Chen JY. Development of reduced mechanisms for numerical modelling of turbulent combustion. Presented at Workshop on Numerical Aspects of Reduction in Chemical Kinetics, CERMICS- ENPC, Cite Descartes, Champus sur Marne, France; [24] Glarborg P, Dam-Johansen K, Kristensen PG, Alzueta M, Rojel H. Low temperature nitrogen chemistry. GRI Report 97/0130, Chicago, IL; [25] Patankar SV. Numerical heat transfer and fluid flow, computational methods in mechanics and thermal sciences. Hemisphere Pub. Co., Washington, DC; 1980.