Abstract
Experimental and numerical investigations were performed for pure pulverized biomass combustion in a 300 kW laboratory swirl burner with a pre-combustion chamber. This work investigated a bluff body at the burner tip and how that affected the combustion characteristics in comparison with a conventional annular orifice burner. The combustion performances were assessed by measuring the temperature distribution in a pre-combustion chamber and furnace, oxygen concentration, and emissions (CO and NOx). Simulations were carried out and validated, providing insight on flow aerodynamics, particle trajectories, species concentrations, and temperature in a pre-combustion chamber and furnace. It was concluded that the bluff body provided a superior performance in terms of flame attachment and combustion efficiency. However, the emissions were high due to the contribution of thermal NOx.
Download PDF
Full Article
Development of Pulverized Biomass Combustion for Industrial Boiler: A Study on Bluff Body Effect
Niwat Suksam and Jarruwat Charoensuk *
Experimental and numerical investigations were performed for pure pulverized biomass combustion in a 300 kW laboratory swirl burner with a pre-combustion chamber. This work investigated a bluff body at the burner tip and how that affected the combustion characteristics in comparison with a conventional annular orifice burner. The combustion performances were assessed by measuring the temperature distribution in a pre-combustion chamber and furnace, oxygen concentration, and emissions (CO and NOx). Simulations were carried out and validated, providing insight on flow aerodynamics, particle trajectories, species concentrations, and temperature in a pre-combustion chamber and furnace. It was concluded that the bluff body provided a superior performance in terms of flame attachment and combustion efficiency. However, the emissions were high due to the contribution of thermal NOx.
Keywords: Pulverized biomass; Swirl burner; Bluff body; Pre-combustion chamber; CFD
Contact information: Department of Mechanical Engineering, Faculty of Engineering, King Mongkut’s Institute of Technology Ladkrabang, Bangkok 10520 Thailand;
* Corresponding author: jarruwat.ch@kmitl.ac.th
INTRODUCTION
Coal consumption has become a debatable issue over recent years because it is a non-renewable resource. The concern regarding the impact of emissions from coal combustion on climate change has increased the interest in biomass as an alternative because it is renewable, considering that it is a carbon-neutral energy resource (Sami et al. 2001). However, biomass has lower calorific value because the composition of biomass has a lower carbon content than coal. Additionally, when converting chemical energy to thermal energy via combustion, its thermal performance should be considered, i.e., its flame stability, emissions, etc.
Pulverized biomass has been adopted for co-firing with pulverized coal in the furnace of an industrial boiler for power generation and process steam. Yi et al. (2013) conducted a study with thermogravimetric analysis (TGA) on a blend of ramie residue and coal. The effects of coal blending ratio from 0 to 30 wt% on performance of cyclone furnace were investigated. Ndibe et al. (2015) had successfully run a 300 kW combustor firing 100% pulverized torrefied spruce, 100% pulverized coal, and 50% blending of these two types of fuel with pre-heated secondary air at 195 C. A study using a drop tube furnace was also carried out by Wang et al. (2015) on coal and coal-biomass blending combustion performance (coal, straw, and wood). Results suggested a positive effect of blending on combustion efficiency and NOx emission. The effect of air staging was also evaluated. Aziz et al. (2016) had adopted computational fluid dynamics (CFD) for simulation of combustion performance of pulverized fuel blends between palm kernel and coal in an existing power plant. In their study up to 15% of the biomass could be adopted for co-firing without adverse effects on temperature distribution. Darmawan et al. (2017) performed experimental and numerical investigation on combustion of hydrothermally treated empty fruit bunch (HT-EFB) blended with coal in a drop tube furnace. A feasibility study was also carried out to integrate fuel processing into power plant system. Pu et al. (2017) recently attempted to use coal blended with biomass as a feedstock for an oxy-fuel bubbling fluidized bed combustor.
The existing plants have not been designed for firing pure pulverized biomass. This is due to the low heating value of volatiles as compared with that from coal, leading to problems with its flame stability. Char burnout was relatively slower due to both a larger biomass particle size and slower reaction kinetics. Firing pure pulverized biomass has been tested in the laboratories of some research institutes.
Ballester et al. (2005) carried out experiments on pulverized combustion in a semi-industrial scale furnace with three types of fuel: bituminous coal, lignite, and oak sawdust. Due to the difference in the stoichiometric air ratio of sawdust from coal, the operating conditions for bituminous and lignite were similar; however, there was a significant difference observed for sawdust. The aim of that study was to investigate the effect of different fuel types on some important flame characteristics, such as visible flame shape, distributions of temperature, and some major species (i.e., O2, CO, NOx unburned hydrocarbon, and N2O). It was found that a high volatiles content of lignite led to more intense combustion near the burner when compared with that of the bituminous flame. Such effects were even more prominent in the biomass flame due to it having the highest value of volatiles content. In addition, the results of hydrocarbon concentration and CO concentration confirmed this finding. Therefore, an expanded combustion zone of biomass had been found. However, according to the percentage of unburned carbon 1 m from the burner outlet, the combustion of bituminous particle was the lowest, followed by lignite, and the biomass particles gave the highest percentage of unburned carbon. There were two distinct regions of combustion, with the fist zone exhibiting intense volatiles combustion, followed by the region with trace amounts of delayed volatiles release and the burning of solid char.
Computational fluid dynamics has been widely used for the research and development of a pulverized combustion system. Ma et al. (2007) simulated the combustion of biomass in an existing pulverized coal-fired furnace by using a developed CFD model. The prediction showed a reasonable agreement with experimental data. Yin et al. (2012) investigated the combustion characteristics when firing pure coal and firing pure wheat straw in a 150 kW swirl-stabilized burner flow reactor under nearly identical conditions. Their results showed dissimilar combustion characteristics between the coal flame and the straw flame. Li et al. (2013) investigated the combustion characteristics of pulverized torrefied-biomass firing with high-temperature air by experiment and by simulation using CFD. They reported on the effect of drag force on the devolatilization of biomass and on the flame characteristics. The effect of oxygen concentration in an oxidizer and of air velocity on the flame characteristics were additionally discussed.
Elfasakhany et al. (2013) studied the combustion of pulverized biomass by conducting experimental trials and performing a simulation with CFD. There were three groups of particle size distributions ranging from 0.02 mm to 1.2 mm with an aspect ratio greater than 5. The study found a high volatiles release with high concentrations of hydrocarbon and carbon monoxide. The flame resembled a gaseous premixed flame. Three distinctive zones are identified: the preheated zone, the devolatilization zone, and the char oxidation zone. The heat transfer mechanism of the preheated zone for biomass was different from that of the gaseous combustion because the biomass particle was heated up by the radiation mode of heat transfer from the flame and furnace wall prior to undergoing the process of devolatilization. Therefore, pollutant emissions were affected by the burner configuration, particle size, and the furnace chamber. In addition, there was a “rocket effect” during the devolatilization process of the biomass particles.
Weber et al. (2015) studied the variety of combustion behaviors in mixed wood, sawdust, fermentation-process residues and grain residues, and South African Middleburg coal. The pulverized coal was injected through an annular nozzle surrounded by the combustion air without swirl, and the biomass was injected through a central pipe nozzle. The furnace was cylindrical with a vertical orientation. The firing rate was 15 kW for all cases. Important flame characteristics such as ignition, temperature rise and distribution, devolatilization, NOxemission, combustible burnout, fly ash, and slag were investigated. The peak temperature in the volatiles combustion zone when firing coal was 100 C higher than that of biomass, with an earlier ignition observed for coal. However, all types of fuel fired under this experimental setting achieved a burnout greater than 94.2%. Approximately 50% to 60% of the nitrogen content in sawdust and mixed wood was converted to NOx, while 16% to 18% was converted for the coal, fermentation-process residues, and grain residues. The bottom ash deposition was found at 950 C to 1200 C, and the deposition for the biomass combustion was three times greater than that of the coal combustion.
Karim and Naser (2018) developed user-defined subroutines on a commercial CFD code (AVL Fire) for simulation of woody biomass combustion on moving grate boiler. They found an agreement between the simulation result and the experimental data. Gómez et al. (2019) developed a mathematical model for porous media. The model was used together with a commercial software, FLUENT, to simulate combustion, heat and mass transfer in a 35 kW wood pellet furnace. The effect of exhaust gas recirculation (EGR) and excess air was investigated. The model had reasonably represented the phenomena taking place in the reactor.
Elorf and Sarh (2019) found the effect of excess air on the flow aerodynamics and the combustion characteristics of pulverized biomass to be made of olive cake in a vertical furnace with a cylindrical cross section. Fuel was injected radially from the bottom of the furnace to increase the residence time. The study adopted CFD to perform a 3D reacting flow simulation. Biomass with an average particle size of 70 µm was injected radially and combusted under excess air ratios of 1.3, 1.7, 2.3, and 2.7. The simulation suggests that the increase in the excess air ratio would dilute the CO and CO2 concentrations, decrease the flue gas temperature, and shorten the flame length.
Combustion of biomass in pulverized form has the potential for replacement of conventional fossil fuel. In a conventional burner configuration, fuel is fed into the combustion chamber by pneumatic transport surrounded by combustion air. Introduction of swirl motion to the primary or secondary air stream would help spreading the solid particle around, leading to rapid heat transfer as well as the enhancement of devolatilization and its reaction rate (Zhou et al. 2003; Gu et al. 2005). However, when combusting biomass with different ignition and char burnout characteristics, additional modification is necessary to improve the combustion performance of the burner.
This paper reports on an achievement made on combustion of pure pulverized biomass with a new burner specifically designed and constructed for pure biomass combustion. The burner is equipped with a pre-combustion chamber, which is used to accommodate ignition of biomass volatile compounds. Apart from the result of the base case, development work is also carried out by installing a bluff body at the burner tip locating at the center of primary air inlet. Discussion is made on how this bluff body affects to the aerodynamics at the central region of a pre-combustion chamber, which leads to improvement of combustion stability and efficiency. The result is verified by temperature distribution and CO emission taken from experiment performed by the authors. CFD is used to investigate aerodynamics, particle trajectories, devolatilization, and combustion when the burner is operated with and without bluff body. Species concentration and NOx emission are also reported. This simulation enable an in-depth understanding on how the bluff body affects to aerodynamics in the near burner region and eventually to combustion stability and emissions.
EXPERIMENTAL
Experimental Setup
The pulverized biomass swirl burner assembled with a pre-combustion chamber had diameters of 0.385 m and 0.465 m. The furnace had 1 m in inner diameter and 1.215 m in length, as shown in Figs. 1 and 2.
Fig. 1. Pulverized biomass combustion system: (1) pulverized biomass milling and feeding system, (2) primary air and fuel inlet, (3) secondary air inlet, (4) tertiary air inlet, (5) swirl burner and pre-combustion chamber, (6) furnace, (7) flue gas treatment and wet scrubber, (8) air supply system, (9) ID fan, and (10) flue gas stack
Fig. 2. Pulverized biomass combustion burner and furnace
The burner was operated at 300 kW thermal throughput (net) at an equivalence ratio of 0.834 (20% excess air). The fuel and air entered the furnace at 313 K. Primary air and pulverized biomass entered through a central pipe with diameter of 0.0525 m. The primary air and fuel were fed at 0.0186 kg/s and 0.0134 kg/s, respectively. Secondary air travelled through swirl vanes, creating a swirl flow, with a swirl number equal to 0.83, through the annular pipe before entering the pre-combustion chamber at 0.0869 kg/s. Tertiary air entered the furnace from the connector between the pre-combustion chamber and furnace at 0.0334 kg/s, as presented in Table 1. The induced draft (ID) fan was used to draw flue gas from the furnace to the stack. The vacuum outlet of flue gas was controlled at 1500 Pa below atmospheric pressure. The flame temperature was measured using a type K thermocouple with a Yokogawa XL100 Portable Data Station (Yokogawa Electric Corporation, Tokyo, Japan). The combustion gas species were measured using a Testo 330-2 LL flue gas analyzer (Testo AG, Lenzkirch, Germany). The monitoring locations of the pulverized biomass furnace are illustrated in Fig. 3.
Table 1. Operating Conditions of the Furnace
Fig. 3. The monitoring locations of the pulverized biomass furnace. The locations of T1, T2, T3, T4, T5, T6, T7, T8, and T9 have distances from the burner exit as follows: 0.082 m, 0.182 m, 0.282 m, 0.382 m, 0.482 m, 1.081 m, 1.594 m, 2.107 m, and 2.681 m, respectively.
Bluff Body
Two types of nozzles were used. The annular nozzle in Fig. 4a was equipped with a liquefied petroleum gas (LPG) burner at the center of the pipe, resulting in an area ratio of 0.23. A later version (Fig. 4b) was equipped with a 6 mm bluff body with a 45 degree cone angle detached from the LPG burner. This later version provided the blockage ratio of 0.42.
(a)
(b)
Fig. 4. The two cases of the burner tip in the pre-combustion chamber, with (a) being the annular orfice and (b) being the bluff body
The definitions for the area ratio and blockage ratio (b.r.) were similar. It was defined as the cross-sectional area of the no-flow section to the total cross-sectional area of a primary air pipe, as shown in Fig. 4 and Eq. 1,
(1)
where d is the bluff body diameter (m), and D is the primary air pipe diameter (m).
Pulverized Biomass
The pulverized biomass was made from a rubber tree from the southern part of Thailand. Initially, it was pelletized and packed prior to shipment. It arrived at the test site, was pulverized by a hammer mill, and was filtered by a 0.5 mm perforated stainless plate (Fig. 5). The size distribution of the particle after passing through the perforated plate is given in Table 2. The proximate and ultimate analyses of the pulverized biomass are given in Table 3.
Fig. 5. Pulverized biomass fuel, (a) pellet biomass fuel from rubber tree before milling and (b) pulverized biomass fuel after milling
Table 2. Pulverized Biomass Sieve Analysis
Table 3. Proximate and Ultimate Analysis of Pulverized Biomass Used
HHV: higher heating value; LHV: lower heating value
Mathematical Modeling
The CFD software FLUENT was used to solve the discretized equation of fluid flow combustion, heat and mass transfer, and the tracking of a solid particle. The computational domain of the pulverized biomass combustion furnace is shown in Fig. 6. The three-dimensional computational domain with a hybrid mesh made with both hexahedral and polyhedral mesh was used. There were approximately 600,000 cells for this simulation. The standard model with the standard wall function was used to model the turbulent flow (Launder and Spalding 1974). The radiative heat transfer included in the simulation was the discrete ordinates (DO) model (Raithby and Chui 1990; Chui and Raithby 1993; Murthy and Mathur 1998). The absorption coefficient for radiation was calculated using the weighted sum of gray gas model (WSGGM) (Smith et al. 1982; Coppalle and Vervisch 1983).
Fig. 6. Computational domain of the pre-combustion chamber and furnace
Discrete phase model
To simulate the pulverized biomass combustion, the discrete phase model (DPM) was adopted using the Eulerian-Lagrangian method. The gas phase was modeled by the Eulerian method, while the discrete solid phase of biomass particles was modeled by the Lagrangian method.
Combustion model
The single kinetic rate model was employed to predict the devolatilization of pulverized biomass (Badzioch and Hawksley 1970). The homogeneous combustion of gas phase was predicted using eddy-dissipation model (Magnussen and Hjertager 1977). The two-step global reaction of the gas phase is modeled in Eqs. 2 and 3.
1.04C 2.32H 1.04O 0.013N 0.58O2 1.04CO 1.16H2O 0.0065N2 (2)
CO 0.5O2 CO2 (3)
The char oxidation was determined by the kinetic/diffusion-limited rate model (Field 1969; Baum and Street 1971). The biomass char surface oxidation model is shown in Eq. 4.
C(s) 0.5O2 CO (4)
Nitric oxides model
In the present work, the nitric oxide (NOx) model consists of NOx formation from thermal NOx, prompt NOx, and fuel NOx. In addition, the NOx reduction from the reburn mechanism was included in the simulation. Thermal NOx refers to the NOx formed via the high temperature oxidation of the nitrogen of air in the combustion. The rate of thermal NOx is calculated by the extended Zeldovich mechanism (Zeldovich et al. 1947). The prompt NOx is generated by the reaction between nitrogen and hydrocarbon in a fuel-rich zone of combustion (Fenimore 1971). Fuel NOx is formed by the reaction between the oxygen and nitrogen contained in the fuel (De Soete 1975). The NOx reburn mechanism was modeled via the reaction of NOx with HCN to form N2. Another reaction pathway is between the NOx and CHi radicals that formed HCN, which eventually reacts with NOx via the former reaction (Kandamby et al. 1996).
Numerical method
The solution methods used for the numerical simulation were the semi-implicit method for pressure linked equations (SIMPLE) for the pressure-velocity coupling scheme, the spatial discretization used the least-squares cell based for gradients, the Pressure Staggering Option (PRESTO!) for pressure, and second order upwind for what remained, i.e., momentum, turbulent kinetic energy, turbulent dissipation rate, gas species, energy, and DO for radiation heat transfer.
RESULTS AND DISCUSSION
Experimental Results
Temperature distribution
Figure 7a shows the temperature distribution in the combustion chamber of the base case having an annular outlet with an area ratio of 0.23. The flame was visualized at a certain distance from the burner outlet, next to the cloud of unburned particles. In contrast, Fig. 7b shows the cone-shaped bluff body with a blockage ratio 0.42, and the suggested attached flame, with a strong illumination of the flame, next to the burner outlet. A relatively greater degree of particle dispersion was observed, implying that the particle could receive more heat transfer from the convection of hot gas and radiation from the refractory of a pre-combustion chamber. In comparison with the annular orifice case, a relatively greater wake region was created next to the bluff body, which was an additional contribution to this effect. These observations agreed with the findings of Liu et al. (2016).
Figure 8 shows the axial distribution in temperature. This figure supports the discussion regarding flame attachment as given in an earlier paragraph. The case with a blockage ratio 0.42 provided higher temperatures next to the burner inlet, especially at T2 where the temperature was 497 K higher than that of the same location for the annular orifice case. This indicates there was an earlier ignition. In addition, the furnace temperature was 145 K higher than that of the annular orifice case, implying a higher char reactivity. However, there was small difference in the flue gas temperature at the furnace exit (T9), with a temperature that was 79 K higher than the annular orifice case.
Species concentration
Figure 9 shows the oxygen (O2) concentration at the furnace exit. The annular orifice case had a greater amount of O2 left unconsumed, which is logically related to an earlier observation on temperature distribution from Fig. 8.
The carbon monoxide (CO) concentration at the exit plane of the furnace was compared between the base case and the case with a higher blockage ratio, as shown in Fig. 10. The annular orifice had a higher concentration of CO, suggesting there was a lower proportion that burned out. This suggested that the annular orifice nozzle had allowed too much penetration of the primary jet, causing a significant amount of unburned particle to pass through the pre-combustion chamber. These particles when entering the main chamber were quenched by relatively cooler environment. This was unlike the case equipped with a bluff body of blockage ratio 0.42, where a greater portion of the fuel particles had enough residence time to burn in the pre-combustion chamber. This latter case yielded less CO emission at the furnace exit plane.
(a)
(b)
Fig. 7. Flame inside the pre-combustion chamber of the annular orifice (a) and bluff body case (b)
Fig. 8. Axial temperature distribution inside the pre-combustion chamber and furnace
Fig. 9. Oxygen (O2) concentration (vol%, dry) at flue gas outlet
Fig. 10. Carbon monoxide (CO) concentration (ppm, dry) at flue gas outlet
NOx concentration
Figure 11 shows the comparison of NOx concentration at the furnace exit. The case with the bluff body with more fuel burn out and a higher flue gas temperature yielded higher NOxemissions as compared with the annular orifice case. This relationship suggests that NOx could be attributed to thermal NOx via the Zeldovich mechanism. Although NOx could come from the reaction involving the nitrogen content in fuel (fuel NOx), between OH radicals and nitrogen molecules near the flame (prompt NOx), or other formation pathways, i.e., intermediate N2O, etc., their contributions to the total NOx emissions were considered small. For instance, the nitrogen content in fuel accounted for 0.49 wt%, dry ash-free. However, the difference in the NOx emissions for the annular orifice case was twice as much as the cone-shaped bluff body case.
Fig. 11. Nitric oxide concentration (ppm, dry) at flue gas outlet
Model Validation
To obtain a sensible prediction of the numerical results, the model was validated against the experimental results of the bluff body case. Comparisons between the CFD data and the experimental data on the axial temperature profile and species concentrations are shown in Figs. 12 through 15.
Fig. 12. Comparing axial temperature between CFD and experimental results for the bluff body
Fig. 13. Comparison of O2 concentration (vol%, dry) between the CFD and experimental (EXP) results at the flue gas outlet for the bluff-body case
Fig. 14. Comparison of the CO concentration (ppm, dry) between the CFD and experimental (EXP) results at the flue gas outlet for the bluff-body case
Fig. 15. Comparison of nitric oxide (NOx) concentration (ppm, dry) between the CFD and experimental (EXP) results at the flue gas outlet for the bluff-body case
The results were generally in agreement because both the experiment and prediction give resembled trends along the axial direction from inlet to the furnace exit plane. However, when focusing on certain locations (e.g., T2, T3, and T4), the prediction provided a slight under-estimation of the flue gas temperature. Nevertheless, the effect caused by the difference in the inlet configuration was reflected in the prediction of the flue gas temperature at T1. The prediction of oxygen concentration at the furnace exit plane was similar to the measured result. However, the model under-predicted the CO concentration when compared with the experimental data. This was due to the fact that the prediction of CO oxidation rate in gaseous phase was defined by the eddy dissipation model, which is based on an assumption of fast kinetics. Development of a CO oxidation model for post combustion zone under an oxygen-depleted environment is an interesting topic of future research, as the reaction rate of CO is more likely to be governed by chemical kinetics. As for the NOx emissions, the prediction agreed with the experimental results.
Numerical Results
Flow fields
Figure 16 shows the predicted axial velocity component (x-velocity) inside the pre-combustion chamber and furnace. As expected, the case with the annular orifice had a higher magnitude of x-velocity. In the wake region, next to the bluff body, the magnitude of the negative velocity was greater than that of the annular orifice. Moreover, the region with the negative x-velocity value was larger than the annular orifice case. The maximum predicted value in this region was as low as -4.33 m/s for the bluff body case. This effect helped bring the hot flue gas that was downstream of the pre-combustion chamber backward to the region next to the burner exit and promoted volatiles ignition in an “internal recirculation zone.” An increase in the flue gas temperature in the pre-combustion chamber caused an increase in the specific volume, thus resulting in a higher flue gas velocity at the connecting port between the pre-combustion chamber section and the main furnace.
Fig. 16. Axial velocity magnitude (m/s) inside the pre-combustion chamber and furnace for the annular orifice case (a) and bluff body case (b)
Figure 17 illustrates the velocity vector and its magnitude in the pre-combustion chamber. For the annular orifice case (Fig. 17a), the directions of the velocity vectors around the centerline pointed toward the central region of the pre-combustion chamber. However, the external recirculation zone had a relatively lower magnitude as compared with those found in the case with the nozzle equipped with the bluff body (Fig. 17b). In the latter case, a stronger reverse flow, indicated by the direction of the velocity vectors, pointed backward to the bluff body. In addition, the region of the reverse flow was greater than that of the annular orifice case.
Fig. 17. Velocity vectors colored by the velocity magnitude (m/s) inside the pre-combustion chamber for the annular orifice case (a) and the bluff body case (b)
The particle trajectories of the biomass for both cases are shown in Fig. 18, with different color shades indicating the different levels of volatiles content in the particles. The prediction gives a clear explanation for the experimental data because there was less particle dispersion for the annular orifice case when compared with the bluff body case. This resulted in a relatively poor release of volatiles. For the annular orifice case, volatiles had not completely released from the particle when leaving the pre-combustion chamber (Fig. 18a), while almost all the volatiles had released for the bluff body case. The asymmetry of all the species’ concentrations and temperatures discussed in the following sections were due to a gravitational force that brought the particle down toward the lower part of the chamber.
Temperature field
The temperature distribution in Fig. 19 suggests that the bluff body created a high temperature next to it, while a lower temperature value was observed with the annular nozzle. The effect of the aerodynamics that led to this observation was discussed earlier in the experimental results. However, the distribution in the furnace next to the combustion chamber was relatively similar.