当前位置:首页 >> 能源/化工 >>

Low Oxygen Dilution (MILD) Combustion in a Recuperative Furnace


Energy Fuels 2009, 23, 5349–5356 Published on Web 10/06/2009

: DOI:10.1021/ef900866v

Importance of Initial Momentum Rate and Air-Fuel Premixing on Moderate or Intense Lo

w Oxygen Dilution (MILD) Combustion in a Recuperative Furnace
Jianchun Mi,*,?,§ Pengfei Li,? Bassam B. Dally,? and Richard A. Craig?
?

Department of Energy & Resources Engineering, College of Engineering, Peking University, Beijing 100871, People’s Republic of China, ?School of Mechanical Engineering, The University of Adelaide, South Australia, 5005, Australia, and §State Key Laboratory of Coal Combustion, Huazhong University of Science and Technology, Wuhan 430074, People’s Republic of China Received May 25, 2009. Revised Manuscript Received September 11, 2009

This paper reports an investigation on the influences of air-fuel injection momentum rate and the air-fuel premixing on the moderate or intense low oxygen dilution (MILD) combustion in a 20-kW recuperative furnace. Various patterns of partially and fully premixed reactants have proven experimentally to work extremely well in the present furnace. H2 recorded for a variety of equivalence ratios at a firing rate of 10 kW. The present numerical study suggests that there is a critical momentum rate of the inlet fuel-air mixture below which the MILD combustion cannot occur. Also, it is revealed, both experimentally and numerically, that, above the critical rate, both the inlet fuel-air mixedness and momentum rate impose insignificant influence on the stability of and emissions from the MILD combustion.

1. Introduction Combustion of a mixture of reactants diluted with exhaust gases, at a temperature above that of autoignition, can achieve reduced NOx emissions and enhanced thermal efficiency.1,2 The application of this combustion principle to practical systems has taken different routes, and different names are used to describe the process. Some relied on a descriptive form of the resulting combustion process (i.e., flameless oxidation), whereas others described the features of the reactants streams (i.e., high-temperature air combustion). The term used in this paper is moderate or intense low oxygen dilution (MILD) combustion.3 A good volume of work has been published on MILD combustion, which was reviewed in a book.4 Gaseous, liquid, and solid fuels were investigated, and some commercial products have been available for a while in the market.5,6 Further application of this mode of combustion to wasteto-energy technologies is also underway.7 On the fundamental side, Cavaliere and de Joannon,3 in a review paper, highlighted the current knowledge in the field and emphasized two main points. The first is that MILD
*Author to whom correspondence should be addressed. Phone: ?86 10 6276 7074. Fax: ?86 10 6276 7074. E-mail: jcmi@coe.pku.edu.cn. (1) Wunning, J. A.; Wunning, J. G. Prog. Energy Combust. Sci. 1997, 23, 81–94. (2) Katsuki, M.; Hasegawa, T. Symp. (Int.) Combust. 1998, 27, 3135-3146. (3) Cavaliere, A.; de Joannon, M. Prog. Energy Combust. Sci. 2004, 30, 329–366. (4) Tsuji, H.; Gupta, A. K.; Haskgawa, T.; Katsuki, M.; Kishimoto, K.; Morita, M. High Temperature Air Combustion;From Energy Conservation to Pollution Reduction; CRC Press: Boca Raton, FL, 2003. (5) Blasiak, W.; Yang, W. Volumetric Combustion of Coal and Biomass in Boilers. In Proceedings of the HITAC Conference, Phuket, Thailand, 2007. (6) Blarino, L., Fantuzzi, M.; Malfa E., Zanusso, U., Tenova Flexytech burners: Flameless Combustion for very low NOx Reheating Furnaces. In Proceedings of the HITAC Conference, Phuket, Thailand, 2007. (7) Yoshikawa, K. R&D Commercialization of Innovative Waste-toEnergy Technologies. In Proceedings of the HITAC Conference, Phuket, Thailand, 2007.
r 2009 American Chemical Society

combustion is different from other combustion regimes and should be treated as such, and the second is that MILD combustion applies for a narrow range of temperature that allows design, optimization, and adjustment in the process by fine-tuning external parameters. On a separate study, de Joannon et al.8 conducted zerodimensional analysis of methane-diluted oxidation using the PSR code of the Chemkin package. This approach was meant to study the theoretical limits of practical flameless oxidation processes. They found that flameless oxidation could be schematized as a two-stage process, the first of which is fuelrich followed by a second stage, which involves a diluted reaction process. Dally et al.9,10 investigated the detailed structure of hydrocarbon jet flames issuing into a hot and diluted coflow to emulate MILD conditions. They used advanced laser diagnostics techniques to evaluate the interaction of turbulence and chemistry and provided valuable experimental data for modellers of these flames.11 Medwell et al.12,13 extended this work to investigate the effects of fuel type on flame stability and lift-off height and provided an insight on the evolution of the OH radical and the formaldehyde intermediate to better understand the autoignition phenomenon prevalent in these flames. On the application side, Dally et al.14 used a recuperative MILD combustion furnace to investigate the effect of fuel
(8) de Joannon, M.; Saponaro, A.; Cavaliere, A. Proc. Combust. Inst. 2000, 28, 1639-1646. (9) Dally, B.; Karpetis, A. N.; Barlow, R. S. Proc. Combust. Inst. 2002, 29, 1147-1154. (10) Dally, B. B.; Karpetis, A. N.; Barlow, R. S. Australian Symposium on Combustion and the Seventh Australian Flame Days, Jan. 2002, Adelaide, Australia. (11) Christo, F. C.; Dally, B. B. Combust. Flame 2005, 142, 117–129. (12) Medwell, P. R.; Kalt, P. A. M.; Dally, B. B. Combust. Flame 2007, 148, 48–61. (13) Medwell, P. R.; Kalt, P. A. M.; Dally, B. B. Combust. Flame 2008, 152, 100–113. (14) Dally, B. B.; Riesmeier, E.; Peters, N. Combust. Flame 2004, 137, 418–431.

5349

pubs.acs.org/EF

Energy Fuels 2009, 23, 5349–5356

: DOI:10.1021/ef900866v

Mi et al.

type on the stability and characteristics of the flames. They found that premixing the fuel with an inert gas, which simulates exhaust gas recirculation to the fuel jet, helps achieve the MILD conditions without the need for high jet momentum. They also found, through computational studies, that the dilution with inert gas helps shift the stoichiometric contour to the high scalar dissipation region within the jet, which suppresses flame propagation and leads to a distributed et al.15,16 designed and reaction further downstream. Szego tested the MILD combustion furnace (MCF) for gaseous fuels. They reported measurements of temperature and fluegas composition from conventional and MILD flames using compressed natural gas (CNG) and liquefied petroleum gas (LPG) as fuels. In their setup, the fuel in the MILD mode was injected through four small jets, which were located away from the center and close to the furnace walls, while the air was injected through the central nozzle. They investigated the NOx emission from a 20-kW MILD combustion furnace. They found that air preheating is not required to achieve MILD combustion, even with 40% of useful heat being extracted through the cooling loop. Based on 191 different measurement cases, covering different fuels and dilution levels, they also found that neither kinetics nor mixing controls the scaling of NOx emissions. They concluded that the effects of both mixing and chemistry must be included in any model to predict NOx emissions in the MILD regime accurately. Szego et al.17 investigated the stability and operational condition for a variety of fuels and jet momentums, using the same furnace. They found that, for the geometry they used, a certain fueljet momentum threshold was needed to achieve MILD conditions. This momentum ensured the penetration of the fuel jets to a region classified as the oxidation zone. They concluded that the parallel jet concept works well when the furnace contains an upper recirculation and mixing zone, which ensures the mixing of the reactants and the combustion products. Weber et al.18 reported on experimental measurements inside a furnace operating with highly preheated air regime. Their data have shown that a substantial improvement (up to 60%) in net flux of the thermal radiation can be achieved under this combustion regime. Both the mixing pattern and intensity have significant effects on the overall performance of the furnace, specifically on the thermal efficiency part. They also noted that further research is still needed to optimize the burner designs to maximize the circulation and mixing inside the furnace. Weber et al.19 later investigated the combustion of light and heavy fuel oils in high-temperature air. They reported excessive soot and pollutants emission and concluded that special atomizers should be developed for the process. Blasiak and Yang5 reviewed the latest developments in using MILD combustion for oxy-fuels, coal under volumetric combustion, and cofiring of coal with biomass. In particular, they investigated the advantages of using rotating opposed
, G.; Dally, B. B.; Nathan, G. J. Combust. Flame 2008, 154, (15) Szego 281–295. , G.; Dally, B.; Nathan, G.; Christo, F. The Sixth Asia(16) Szego Pacific Conference on Combustion, Nagoya, Japan, May 20-23, 2007; pp 231-234. , G.; Dally, B.; Nathan, G. J. Combust. Flame 2009, 156, (17) Szego 429–438. (18) Weber, R.; Orsino, S.; Lallemant, N.; Verlaan, A. Symp. (Int.) Combust. 28, 2000, 1315-1321. (19) Weber, R.; Smart, J. P.; vd Kamp W. Proc. Combust. Inst. 30, 2005, 2623-2629.

Figure 1. Schematic diagram of MILD combustion furnace and parallel jet burner system.

fired air (ROFA) to enhance mixing and internal exhaust gas recirculation in boilers. They found that, by incorporating the high momentum jets, through the ROFA system, it was possible to mix different ratios of coal and biomass and run the boiler with 100% biomass. They also reported a reduction in NOx and CO for all fuel mixtures used. Wang et al.20 reported on a computational study that compares the emission and performance characteristics of the flameless oxidation (FLOX) and continuous stage air combustion (COSTAIR) technologies applied to coal and biomass (straw) fuels. The technologies were applied to a 12 MWe coal-fired, CFPC power plant using the ECLIPSE simulation software. They found that both technologies have the potential to reduce NOx emission by up to 90% and that, although straw emits less pollutants, it has a slightly lower efficiency than coal-fired plants. The objective of the present study is 2-fold, i.e., (1) to investigate via CFD the effect of the inlet air-fuel momentum rate on the combustion mode and (2) to examine experimentally the premixing pattern influence on the stability of and emission from MILD combustion. This paper reports on both experimental and numerical results. 2. Experimental Details
The present study uses a laboratory-scale MILD combustion furnace (MCF), as shown in Figure 1. The furnace has been described in detail elsewhere,15-17 and only a brief description is given here. The combustion chamber is well-insulated with four layers of 38-mm-thick high-temperature ceramic fiber boards, which reduced heat loss from the walls to ?20% of the total heat input. This assists with the establishment and stability of the MILD regime and results in a warm-up time of ?3 h from a cold state to steady-state operation. The MCF has optical access through five openings, which are equally spaced, vertically, down three sides of the furnace. These openings can accommodate interchangeable insulating window plugs or ultraviolet
(20) Wang, Y. D.; McIlveen-Wright, D.; Huang, Y.; Hewitt, N.; Eames, P.; Rezvani, S.; McMullan, J.; Roskilly, A. P. Fuel 2007, 86, 2101–2108.

5350

Energy Fuels 2009, 23, 5349–5356

: DOI:10.1021/ef900866v
fuel nozzle retraction (mm) 0 0 20 40 60 90

Mi et al. Table 1. Details of Six Cases for Modelling the Furnace Flow
case 1 (premixed) 2 3 4 5 6 inlet jet momentum (kg m/s2) 0.4643 0.1018 0.0873 0.0468 0.0360 0.0341

(UV)-grade fused-silica windows. Two U-shaped cooling tubes with variable heat exchange areas are used to control the heat load. The heat exchangers can be inserted through any of the window openings; however, for this investigation, they were positioned in windows A3 and C3, as shown in Figure 1. The furnace was designed for a maximum capacity of 20 kW from the fuel and 3.3 kW from air preheat. A turndown ratio of 1:3 was achieved. The burner design consists of a single fuel tube located on the axis of the furnace, an air annular channel around the tube, and four exhaust ports arranged symmetrically in a ring pattern on the same wall as seen in Figure 1. To investigate the effect of premixing fuel and air, the premixing process was achieved in two different ways. The first, for partial mixedness, involved the retraction of the central fuel jet, which is surrounded by a bluff body, inside a larger tube, thus forming an annulus air jet. The bluff body has a diameter of 18 mm, whereas the outer tube has a diameter of 26.6 mm. In this case, the bluff body was lowered at nine retraction distances from the furnace bottom, varying from 10 mm to 90 mm (i.e., z [mm]= -10 to -90), to allow the fuel and the surrounding air to premix before emerging into the furnace. As the jet is retracted, the air and fuel are mixed to a varied extent. Alternatively, the fuel and air were premixed outside the furnace and introduced through a central jet into the furnace. The premixing means that the fuel and air are emerging together into the combustion chamber of the furnace at higher momentum, so that sufficient hot products are entrained before the oxidation process. When this happens, the flame temperature is reduced and the oxidation happens in a distributed fashion. The time-averaged furnace reference temperature (Safety TC) was measured with a bare, fine-wire type R (Pt-Pt-13% Rh) thermocouple that had a wire diameter of 254 μm and a bead diameter of 1.2 mm under steady-state conditions at (x= 0, y= 0, z = 542.5 mm). The exhaust temperature was measured with a stainless steel sheath type K (Ni-Cr) thermocouple. Global emission levels of CO, CO2, NO, NO2, O2, and UHC were measured using a TESTO Model 350XL portable gas analyzer. The analyzer measurement accuracies were estimated as follows: [O2] = (0.8% of measured value, [CO] = (10 ppm or 5% of measured value (whichever is the smallest), [NO] = (5 ppm, [NO2]= (5 ppm, [CO2]= (0.3% of measured value, and UHC (CxHy)= (20 ppm. Natural gas (NG), with 91.4% CH4 (volume fraction at 288 K), was used in this study. Two heat exchangers were used in windows A3 and C3, and their combined exposed surface areas were 0.060 m2. These heat exchangers removed, on average, 4.01, 4.46, and 4.93 kW of heat for the firing rates of 7.5, 10, and 15 kW, respectively.

3. Air-Fuel Mixedness and Injection Momentum Rate (Computational Fluid Dynamics, CFD) Computational fluid dynamics (CFD) was used to model the flow inside the furnace described in Figure 1. Six cases, which are listed in Table 1, were modeled, and the initial mixedness and momentum rate were determined. All calculations were performed with identical mass flow rates of NG (91.4% CH4 and 8.6% C2H6 in volume fraction) and air for an equivalence ratio of φ = 0.8 and a total volume flow rate of ?4 ? 10-3 m3/s. The commercial computational software package Fluent 6.321 was used in this study. A full threedimensional (3D) structured grid was used and constructed to have small orthogonality deviations. The final grid had 400 000 cells in total and was verified through a grid-independent study. The outlet boundary conditions of the computational domain were set as pressure boundaries (static gauge pressure= 0). The standard k-ε model with the standard wall
(21) Fluent 6.3 User’s Guide; Fluent, Inc.: Lebanon, NH, 2006.

function was used to model the turbulent flow. The differencing scheme used for the convective term is the third-order QUICK scheme. The SIMPLEC algorithm was used for pressure correction. The solution convergence was determined by two criteria, i.e., ensuring (1) the residuals of the solved equations to drop below specified thresholds set at 10-3 for all variables, while a residual of 10-6 was used for the energy equation and (2) the value of a sensitive property at a critical spatial location to be stable and no longer changing with iterations. Figures 2a-f show contours of the mean velocity magnitude (expressed in units of m/s) in the central xz-plane (y= 0), respectively, for the premixed case and those of different retraction distances. To compare the recirculation strengths of the internal flow for the six cases, Figure 3 displays the mean velocity contours (also expressed in units of m/s) in the xy-plane at z = 500 mm, where positive and negative values represent the flow upward and downward, respectively. Figure 2a shows that, for the same volume flow rates of air and fuel, the single premixed fuel-air jet discharges from a central tube into the furnace, both more energetically and at a significantly higher momentum rate than the other cases. Consequently, the recirculation of the furnace flow is much stronger in case 1, as reflected by the magnitude of the mean velocity (see Figure 3). Therefore, the strongest recirculation of the flue gases occurs in the premixed case. In other words, the furnace flow is most dynamic at the highest momentum rate when the fuel and air are fully premixed. Figures 2 and 3 also clearly show the effect of the retraction on the furnace flow. Figures 2b-f, together with Figures 3b-f, signify that the recirculation weakens as the distance of retraction increases. This means that the initial fuel-air momentum rate decreases as the retraction distance increases. Next, we examine the effect of the retraction on the flow near the burner exit (see Figures 2b-f. For the nonretraction case (z= 0), the air jet from a thin annulus and the fuel jet from the central tube come together to form a small recirculation zone (SRZ) following the central tube exit in the furnace. This recirculation zone results from the existence of the bluff body and, more importantly, from the annulus air jet being much stronger than the central fuel jet. Because of this, as the flow proceeds downstream, the annulus jet contracts first and then spreads out in the near field. The fuel and air mix with each other in the SRZ after they have entered the furnace. As the retraction distance of the central fuel tube increases, the SRZ is confined more by the air tube, so that it becomes longer in size, because the effect of the big velocity difference between the fuel and air jets is enhanced by the confinement of the recirculation. It is anticipated that the greater the retraction of the fuel tube, the more mixing of the fuel and air in the SRZ prior to entering the furnace, i.e., the more premixing occurs. That is, the retraction enhances the premixing of air and fuel. To check the detail of the premixing, contours of the mass fraction of CH4, denoted by [CH4], at the nozzle exit plane (z = 0, see Figure 1 or 2) were obtained for the six cases and
5351

Energy Fuels 2009, 23, 5349–5356

: DOI:10.1021/ef900866v

Mi et al.

Figure 2. Contours of the mean velocity magnitude (expressed in units of m/s) in the central xz-plane (y = 0) for the six cases listed in Table 1. (a) case 1: premixed (z = 0); (b) case 2: no retraction z = 0; (c) case 3: retraction z = -20; (d) case 4: retraction z = -40; (e) case 5: retraction z = -60; and (f) case 6: retraction z = -90. Note that the red color in panel a is used for contours of U g 30 m/s.

Figure 3. Mean velocity contours (expressed in units of m/s) in the xy-plane at z = 500 mm: (a) premixed (z = 0); (b) retraction, z = 0; (c) retraction z= -20; (d) retraction z= -40; (e) retraction z= -60; and (f) retraction z= -90. Contours in the central region shown in pink are positive (upward flow), whereas the remainder are negative (downward flow).

are displayed in Figure 4. Note that the color scale is from 0 (blue) to 0.1 (red) and that the red color applies for [CH4] > 0.1. To quantify the initial fuel-air mixedness, the ratio of the mininum to maximum of [CH4] (i.e., R= [CH4]min/[CH4]max) is used as a measure of the averaged mixedness when the fuel and air enter the furnace. Obviously, R = 1 and 0 refer, respectively, to cases 1 and 2. Figure 5 shows the initial mixedness versus the retraction distance. Also included in the figure is the initial rate of the fuel-air momenR tum,J ? FU 2 dAwhere F is the mixture density, U the velocity, and A the nozzle exit area. Among all the test cases, shown in Figure 5, the values of J and R for case 1 are highest, where fuel and air are initially 100% premixed. For the partially premixed cases, as the distance of retraction increases from 0 mm to 90 mm, the initial mixedness between fuel and air increases rapidly from R= 0 to R= 0.9, whereas their total momentum rate decreases from J/Jpre= 22% to 7.3%, relative to the premixed case.
5352

The previously mentioned six test cases were experimentally investigated using the MCF operating at a firing rate of 10 kW. The measured results with some discussions are provided in section 5. 4. Effect of Initial Air-Fuel Momentum Rate on Combustion Mode (CFD) This investigation was conducted using CFD modeling. Different furnace combustion scenarios were calculated when fuel and air were injected separately through the central fuel nozzle and the annulus air nozzle (refer to Figure 1). These calculations of nonpremixed combustion were made by varying the location of the fuel nozzle exit (Zf) from Zf= 0 mm to Zf = 26.2 mm. This variation enables a decrease of the total momentum rate (J) of the NG and air jets, because of the increased exit area of the annulus nozzle caused by the bluff body, under a constant mass flow rate (FUA) of air. It follows

Energy Fuels 2009, 23, 5349–5356

: DOI:10.1021/ef900866v

Mi et al.

Figure 4. Contours of the mass fraction of CH4 at the exit plane of the annulus. Note that (i) the color scale is 0 (blue) to 0.1 (red) and (ii) the red color is used for all [CH4] > 0.1.

0), and constant temperature boundaries, respectively. Two heat exchangers are defined by the constant temperature boundary conditions and the temperature is specified as being 500 K. The total heat flux (convection and radiation) through the heat exchangers is ?4.5 kW. The standard k-ε model with the standard wall function was used for modeling the turbulent flow. The eddy dissipation concept (EDC) with two-step chemical kinetic mechanisms for CH4 and one-step mechanisms for C2H6 were used. To reduce the computational cost of time integration, the in situ adaptive tabulation (ISAT) model of Pope22 was used. The discrete ordinate (DO) radiation model23 was applied for radiation. The NOx was computed by taking equilibrium conditions for the reaction O2 T 2O.24 The NOx formation rate was given by d?NO? ? 2kNIf ?O?eq ?N2 ? dt where kNIf ?m3 =?kmol s?? ? 1:8 ? 1011 exp? 38370 ? T

Figure 5. Initial rate of the fuel-air momentum (J) and initial fuelair mixedness (R) at different retraction distances of the bluff body. Table 2. Details of Six Cases for Combustion Modelling
case a b c d e f exit location of fuel nozzle, Zf (mm) 0 4 8 12 20 26.2 momentum rate of NG and air, J (kg m/s2) 0.122 0.104 0.089 0.075 0.051 0.036

that the effect of J on the combustion mode can be investigated. Table 2 shows the momentum rates (J) versus the locations (Zf). As the nozzle exit area A is increased, the initial momentum rate (FU2A) of air drops. We have also performed similar calculations for premixed combustions (whose detailed data are not presented herein), i.e., premixing NG and air through (i) the central fuel nozzle and (ii) the annulus air nozzle, and their inlet momentum rates were varied by changing the corresponding exit diameters. All the CFD calculations were performed at a firing rate of 10 kW and with an identical mass flow rate of the NG-air mixture at φ = 0.8 and total volume flow rate of ?4 ? 10-3 m3/s. The inlet temperature is 20 °C (293 K). The mixture was assumed to obey the ideal gas law. Inlet, outlet, and wall boundary conditions of the computational domain were set as constant velocity, constant pressure (static gauge pressure=
5353

A second-order discretization scheme was used to solve all governing equations. The SIMPLE algorithm was applied for pressure correction. It is well-known1,3,4 that the MILD combustion occurs only where there is sufficient recirculated exhaust gas diluting the reactants, so that the local volume concentration of oxygen is much lower than 21%. In the present recuperative furnace configuration, the amount of recirculation of combustion products may be controlled by the inlet momentum rate of the fuel-air mixture jet (see Figures 2 and 3). Table 2 shows that the inlet momentum rate reduces gradually from the case that is described by Figure 2a of J= 0.122 kg m/s2 to the case that is described by Figure 2f of J= 0.036 kg m/s2. Our CFD calculations reveal that this reduction results in a decreasing recirculation of combustion products into the upcoming fuel and air jets, similar to those shown in Figures 2 and 3. Such a variation of the inlet-momentum-dependent recirculation is expected to cause a switching between the MILD and conventional combustion modes. This indeed happens. Figure 6 illustrates temperature contours (given in Kelvin) in the central xz-plane (y = 0) of the furnace for different values of Zf (given in millimeters) and, thus, different J values (expressed in units of kg m/s2) when burning NG at the firing rate of 10 kW. Cases shown in Figures 6a and 6b show the flameless MILD mode, whereas cases described in Figures 6e and 6f correspond to the conventional combustion mode with visible flames in the central region of the furnace. Evidently, the cases described in Figures 6a and 6b have more uniform and lower temperature (?1500 K) distributions than do the cases described by Figures 6c-f (?2100 K). High local temperatures in the central flame region of the conventional combustion promote the formation of thermal-NOx, thus with the NOx emission of up to 20 ppm, while the MILD combustion produces almost-zero emission of NOx.
(22) Pope, S. Computationally efficient implementation of combustion chemistry using in situ adaptive tabulation. Combust. Theory Modell. 1997, 1 (1), 41–63. (23) Chui, E.; Raithby, G. Computation of radiant heat transfer on a nonorthogonal mesh using the finite-volume method. Numer. Heat Transfer, Part B 1993, 23 (3), 269–288. (24) Turns, S. An Introduction to Combustion: Concepts and Applications; McGraw--Hill: New York, 1996.

Energy Fuels 2009, 23, 5349–5356

: DOI:10.1021/ef900866v

Mi et al.

Figure 6. Temperature contours (in Kelvin) on the central xz-plane (y = 0) of the MCF modeled for different locations of the fuel bluff-body nozzle (Zf) and, thus, different inlet air momentum rates (Ja) when burning NG at a firing rate of 10 kW. Zf and Ja are in given in units of mm and kg m/s2, respectively. There is a critical momentum rate, Jcr ≈ 0.097 kg m/s2, below which the MILD combustion does not occur.

Figure 7. Images showing a nonpremixed flame, realized by injecting fuel and air through the central and annulus nozzles, respectively, to heat the MCF from a cold state.

Overall, the above calculations seem to have been verified by experiments. For example, the cases that are described by Figures 6e and 6f apply for the startup of most of our previous MILD tests. Figure 7 shows that a nonpremixed flame stabilized by the bluff body was used to heat the furnace from a cold state. The MILD combustion mode was activated once the furnace reference temperature exceeded self-ignition (800 °C), and then the bluff-body nozzle tip was retracted. Furthermore, the case that is described by Figure 6a is indeed a case of MILD combustion, as demonstrated in section 5. Another point also can be made from Figure 6. Namely, there must be a critical value (Jcr) of the total inlet momentum rate of the fuel and air jets, above which the MILD combustion occurs and below which this combustion is difficult to be established. This critical rate for the present nonpremixed combustion should be near the average of the momentum rates for the cases described by Figures 6b and 6c (i.e., Jcr ≈ 0.097 kg m/s2). The present calculations for premixed combustions (not shown here) have also revealed the switching between the two modes of combustion, which is due to variation of the momentum rate. All calculations suggest that the magnitude of the critical rate is dependent on the furnace and burner configurations. For instance, the aforementioned value of Jcr (?0.097 kg m/s2) is different from those for the premixed combustions, which are Jcr ≈ 0.024 and 0.054 kg m/s2, respectively, for premixing NG and air through (i) the central fuel nozzle and (ii) the annulus air nozzle. Importantly, for J > Jcr, the magnitude of J has little influence on the MILD combustion stability and emissions. This is deduced from the comparison between exhaust emissions for the cases that are described by Figures 6a and 6b, and
5354

Figure 8. Temporal variations of the measured temperatures for the MCF operating on NG at 10 kW.

particularly for the premixed cases. Note that the maximum and minimum momentum rates for modeled premixed MILD combustions differ by a factor of 3-4, but that their exhaust compositions are basically no different. Moreover, the effect of premixing at J > Jcr is not expected to be important to the MILD combustion, as is verified experimentally in the next section. 5. Experimental Results and Discussion
Figure 8 shows the temperature-time history at the bottom and top sections of the furnace, as well as that of the exhaust gases and the combustion air for the MCF operating with NG with a

Energy Fuels 2009, 23, 5349–5356

: DOI:10.1021/ef900866v

Mi et al.

Figure 9. Exhaust gas composition and temperature and furnace temperature for various retractions with a heat input of 10 kW and an equivalence ratio of φ = 0.77.

Figure 10. Measured CO and H2 emissions, with premixing at different equivalence ratios and a firing rate of 10 kW.

capacity of 10 kW. The transition to MILD combustion clearly reduces the temperature difference between the top and bottom of the furnace, from 175 °C to 70 °C, and slightly changes the exhaust gas temperature. Also notable, the reduction in the inlet air temperature, by switching the external heater off, has a minor effect on the furnace temperatures. As demonstrated in Figure 7c, the flameless combustion occurs almost volumetrically in the entire furnace, while the conventional combustion occurs within small regions of the furnace (see Figures 7a and 7b) and causes high local temperature gradients. Figure 9 shows the effect of the jet retraction on the exhaust composition and temperature, as well as the furnace temperature, for an equivalence ratio of φ = 0.77 and a heat input of 10 kW. Variations in the exhaust emissions are seen for every retraction distance. Note that the present use of retraction is aimed at examining the effect of premixing of reactants on the MILD combustion characteristics; in addition, the more we retract the fuel jet, the higher the air-fuel premixedness (see Figures 2 and 3). Equally important, the retraction also causes a decrease in the initial rate of the total fuel-air momentum (see Figure 5). Here, it is worth mentioning that the case of no retraction was correctly modeled by CFD (i.e., the case that is described by Figure 6a and Table 2); namely, the numerical and experimental results are globally consistent. Figure 9 shows that the MILD combustion, for all the cases, delivers very low emissions of CO, H2, and NOx. As the fuel nozzle head is retracted from z= 0 to z= -90 mm, while those of CO and H2 experience a slight change (0-4 ppm), the measured emission of NOx increases from 2 ppm to 26 ppm and then decreases to 19 ppm. Accordingly, the case of no retraction seems to give the best overall performance. For this case, the initial momentum rate (defined as 0.23Rpre) is highest, whereas the mixedness (which is equal to zero) is lowest. However, the combustion performance for the maximum retraction (90 mm) is not the worst, although the corresponding momentum rate (given as 0.0734Jpre) is lowest and the mixedness (0.91) is highest. Therefore, these observations suggest that the variation of the momentum rate caused by the retraction over the present measured range does not have significant influence on the performance of MILD combustion. In other words, this has confirmed the point made early from the CFD data that there must be a critical value of the inlet momentum rate of the fuel and air, above which the momentum rate is not crucial to the realization of MILD combustion. The critical momentum rate for the present furnace and burner configurations is Jcr ≈ 0.024 kg m/s2 for the premixed NG-air reactants from a central tube, well below the minimum value (0.034 kg m/s2) of the momentum rate obtained from the retraction of 90 mm (see Table 1), i.e., case 6, where the initial mixedness is >90% and can be considered to be the case of almost fully premixing air and fuel. The observation also suggests 5355

Figure 11. Measured NOx emissions, with premixing at different equivalence ratios and a firing rate of 10 kW.

that the effect of premixing air and fuel on the MILD combustion characteristics is insignificant. The aforementioned suggestions can be further argued based on the performance of the MILD combustion in the case of premixing air and fuel through the central tube into the furnace. Both the initial fuel-air momentum rate and mixedness for this case (i.e., Jpre and Rpre) are much higher than those for the retraction cases (see Figure 5). The high momentum rate by premixing is thought to enhance the entrainment of the combustion products into the fuel and air jet inside the furnace and thus create a better condition for the MILD combustion to occur. However, because the critical momentum rate (Jcr ≈ 0.024 kg m/s2) is smaller than 0.073Jpre, an increase greater than Jcr in the momentum rate may not be very beneficial for the establishment of MILD combustion. It follows that the benefits of premixing are not expected to improve the performance of the furnace significantly. Indeed, this is the case, as demonstrated by Figures 10 and 11. These figures respectively show the exhaust emissions of CO, H2, and NOx for the cases of the 100% premixing and the retraction of 90 mm, more than 90% premixing, for a firing rate of 10 kW and varying equivalence ratios (φ). Evidently, very low emissions were measured at φ e 1.0 for both cases; note that these emissions et al.15,17 are consistent with the previous measurements of Szego for the similar combustion configuration. It is also shown that the emissions did not vary substantially with φ. Comparison of the two cases indicates that the emissions of CO and H2 are, on average, only slightly higher (within experimental uncertainties), while the NOx emissions are considerably lower, for the 100% premixed injection. The latter observation is consistent with the

Energy Fuels 2009, 23, 5349–5356

: DOI:10.1021/ef900866v

Mi et al.

Figure 12. Measurements of the exhaust and furnace temperatures, with premixing at different equivalence ratios and a firing rate of 10 kW.

corresponding temperatures inside the furnace (Tfurnace), i.e., Tfurnace is ?5-27 °C less for the premixed case (see Figure 12). In this context, a conclusion can be drawn that, when the rate of the fuel-air injection momentum is greater than the critical value, both the initial mixedness and momentum rate impose little influence on the MILD combustion. Nevertheless, the critical momentum rate varies from furnace to furnace, because of different configurations, as revealed by the CFD work (see section 4).

momentum below which moderate or intense low oxygen dilution (MILD) combustion does not occur in a recuperative furnace. Above the critical momentum rate, no significant influence on the MILD combustion results, either from variation of the initial momentum rate or from that of the initial fuel-air mixedness. Such a critical momentum rate varies from furnace to furnace when their configuration is different. The present study has also examined, by experimentation, the influence of the air-fuel injection momentum rate and the air-fuel premixing on the performance of MILD combustion. The idea of partially and fully premixing the reactants before entering the furnace was realized experimentally by retraction of the central fuel tube and by introducing both fuel and air through the same central tube, respectively. It was observed that premixing, either partially or fully, produced very low CO, H2, and NOx species within the exhaust gas. These observations apply for all the initial fuel-air momentum rates of the present tests, which are practically greater than their corresponding critical values. Finally, it is important to note that increasing the initial momentum rate well beyond the critical value is not beneficial and, thus, should be avoided. This is because it will consume more energy for achieving a similar performance of the MILD combustion.
Acknowledgment. J.M. gratefully acknowledges the support for this study from both the Foundation of State Key Laboratory of Coal Combustion and the Ministry of Science & Technology of China through an 985-Project. We would also like to thank the referees who provided insightful comments and criticisms to an earlier version of this paper.

6. Concluding Remarks The present computational fluid dynamics (CFD) work has revealed that there is a critical rate of the initial reactant

5356


相关文章:
更多相关标签: