Environ. Sci. Technol. 2006, 40, 4010-4015
Effect of Solids Concentration Distribution on the Flue Gas Desulfurization Process JIE ZHANG, CHANGFU YOU,* HAIYING QI, CHANGHE CHEN, AND XUCHANG XU Key Laboratory for Thermal Science and Power Engineering of the Ministry of Education, Department of Thermal Engineering, Tsinghua University, Beijing 100084, China
A dry flue gas desulfurization (FGD) process at 600-800 °C was studied in a pilot-scale circulating fluidized bed (CFB) experimental facility. Various fresh sorbent distribution types and internal structures were modeled numerically to investigate their effect on the gas-solid flow and sulfate reaction characteristics. Experimental results show that, after the fresh sorbent supply was stopped, the desulfurization efficiency declined rapidly even though the sorbent recirculation was maintained. Therefore, the fresh sorbent is the main contributor to the desulfurization process and the primary effect of the recirculated sorbent was to evenly distribute the fresh sorbent and to prolong the sorbent particle residence time. The numerical results demonstrate that the desulfurization efficiency varied greatly for the various fresh sorbent bottom injection methods. The desulfurization efficiency of the bottom-even injection method was 1.5 times that of the bottom two-sided injection method. Internal structures effectively improved the fresh sorbent solids concentration distribution and the desulfurization efficiency. Optimized internal structures increased the desulfurization efficiency of the bottom twosided injection method by 46%, so that it was very close to that of the bottom-even injection method with only a 4.6% difference.
1. Introduction The SO2 emissions from coal-fired power plants have caused significant environmental and human health effects. Several types of FGD processes have been developed to reduce SO2 emission. The moderate temperature dry FGD process can achieve high desulfurization efficiency with low water consumption and with the production of dry CaSO4 as the main byproduct, which makes SO2 removal very attractive. Since this moderate temperature desulfurization process is a dry process, the chemical reaction between the sorbent and SO2 is a gas-solid reaction so the ultimate reaction extent depends on the SO2 diffusion into the sorbent particles. Therefore, the desulfurization efficiency not only depends on the sorbent reactivity and the reaction temperature but also on the contact efficiency and the contact residence time between the SO2 and the sorbent particles, which depends on the gas-solid flow field structure. * Corresponding author phone: +86-10-62781740; fax: 86-1062781740; e-mail:
[email protected]. 4010
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 12, 2006
The 600-800 °C temperature window in the moderate temperature dry FGD process effectively enhances the sulfate reaction rate due to the relatively high temperature. The process not only overcomes the need for large volumes of water to increase the sulfate reaction rate in lower temperature desulfurization processes such as the wet and hemidry technologies but also avoids the reduced desulfurization efficiency due to the limited specific area of sintered sorbent particles in a higher temperature desulfurization process such as furnace injection technologies. Furthermore, the temperature window effectively restrains the carbonate reaction between CO2 and the sorbent, reducing the CO2 and SO2 competition reaction and ensuring effective sorbent utilization (1). A new highly efficient calcium-based sorbent has been developed whose calcium conversion rate increases rapidly with increasing temperature at moderate temperatures (2). The calcium conversion rate at 750 °C reached 95.7% after 90 min of sulfate reaction in a thermogravimetric analysis (TGA) reactor (1). However, the calcium conversion rate was much lower in a pilot-scale CFB experimental facility for the same operating parameters as the TGA experiments (3). The main reason for the unrealizable maximum calcium conversion rate may lie in the improper gas-solid flow field in the CFB facility. CFB reactors not only provide sorbent recirculation and long sorbent residence times but also provide intense gas-solid interactions for FGD technologies. However, if the gas-solid flow field is improperly organized, the fast fluidization regime in the CFB reactor can result in sorbent particles concentrating near the wall leading to a very uneven solids concentration distribution. The reduced gas-solids mixing would then reduce the desulfurization efficiency. The uneven solids concentration distribution in CFB reactors has been widely recognized (4-6). The gas-solid flow behavior in the CFB reactor influences the bed pressure drop, gas-solid mixing extent, gas-solid contact efficiency, and chemical reaction characteristics. Therefore, the gassolid flow organization determines the CFB reactor performance. Many researchers have studied various methods to improve the solids concentration distribution uniformity, such as optimized reactor structures, external force disturbances, particle designs, and elevated gas pressures (7). One of the most practical solutions is the addition of internal structures in the CFB reactor. Current research on internal structures has focused on the petrochemical field. The internal structures are mainly bluff bodies and perforated plates which reduce the axial solids mixing, improve radial solid distribution, and make solids residence time more uniform which ensures that the particle reaction rates are similar so as to achieve homogeneous intermediate products (8-12). However, the internal structures in CFB desulfurization equipment are mostly ring shaped with the aim to improve the gas-solid contact efficiency through enhanced axial solids mixing and prolonged solids residence times (13). However, this research is less developed and the hydrodynamics mechanism is still not well understood. Therefore, studies on the effects of various solids concentration distributions and internal structures on the CFB desulfurization processes are needed to realize high desulfurization efficiencies and low operating pressures. The effect of the sorbent solids concentration distribution on the FGD process was studied experimentally and numerically. The experiment results indicated that the fresh sorbent was the main contributor to the desulfurization process and the primary effect of the recirculated sorbent 10.1021/es060665m CCC: $33.50
2006 American Chemical Society Published on Web 05/11/2006
FIGURE 1. Pilot-scale CFB reactor experimental system diagram. was to evenly distribute the fresh sorbent. Various fresh sorbent distribution types and internal structures were consequently modeled to further investigate the effect of the solids concentration distribution on the calculated gas-solid flow and sulfate reaction characteristics. The recirculation of sorbent particles was omitted in the modeling to help exclude the diffusion effect of recirculation sorbent particles and emphasize the solids concentration distribution of the fresh sorbent itself. The numerical results are useful to provide guidance for gas-solid flow field optimization to improve the desulfurization efficiency in CFB reactors.
2. Experimental Section The pilot-scale CFB experimental system is shown in Figure 1. The main subsystems were the sorbent preparation system, the flue gas generation system, and the CFB reactor. Detailed descriptions of the system were given earlier (1-3). Flue gas generated by the oil burner was mixed with a small amount of cool air to produce 600-800 °C simulated flue gas. SO2 was added to the flue gas before the CFB reactor. The CFB reactor riser was 6 m high with a diameter of 0.305 m and a flue gas flow rate of 300 (N m3)/h. The flue gas passed through the CFB reactor and reacted with the sorbent and then went through a two-stage cyclone separator and bag filter before being emitted from the stack. The sorbent particles collected in the cyclone separator and the bag filter were fed back into the reactor for further circulation or drained out of the system. The O2, CO2, and SO2 concentrations in the flue gas were measured on-line at the CFB reactor inlet and outlet. The desulfurization efficiency was directly calculated from the inlet and outlet SO2 concentrations. The sorbent was produced by quick hydration of lime and coal fly ash, using hydration at ambient temperature for about 2 h and drying at 150-300 °C for 0.5-1 h (14). The average diameter of the fly ash was about 73 µm (3). Scanning electron microscopy (SEM) images of a prehydrated fly ash particle and a hydrated sorbent particle are shown in Figure 2. The original smooth surfaces of the fly ash particles became
FIGURE 2. SEM images of fly ash and hydrated sorbent particles (1300× magnification). very rough after hydration, and the many tiny Ca(OH)2 particles produced by the hydration were about 5 µm in diameter (3). They covered the fly ash particle surfaces, forming a cubic and porous surface structure on the new VOL. 40, NO. 12, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4011
sorbent. The microscopic porous structure was much better than that of the original lime. TGA results showed that the sorbent reactivity was 10 times better than the original lime (3, 14).
3. Numerical Modeling The flow in the reactor was assumed to be a viscous, incompressible, steady, isothermal, turbulent gas-solid flow. The standard k- model was used to simulate the turbulent gas flow in the relatively simple gas flow field. The sorbent particle diameter was assumed to be 83 µm. The inner layer was the fly ash with a particle diameter of 73 µm. The outer layer was a 5 µm thick CaO layer, which approximately represents the real sorbent with tiny absorbent particles covering the fly ash particle. The numerical study was to further investigate the importance of the solids concentration distribution of the fresh sorbent. The recirculation of sorbent particles was omitted to exclude the effect of recirculation sorbent particles and emphasize the effect of fresh sorbent distribution itself. In the calculations, the maximum inlet solids concentration is 0.342 kg/m3 and the solid phase volume fraction of solid phase is only 0.01%. The average distance between particles is 1428 µm, and the Stokes number is only 0.03 (15). Therefore, the gas-solid flow is taken to be a dilute gas-solid twophase flow with the interactions between particles neglected. The stochastic particle trajectory model was used to simulate the sorbent particle motion, taking into account turbulent dispersion of the particles. The effect of the random gas velocity fluctuations on the particle motion was considered by integrating the trajectory equations for individual particles, using the instantaneous gas velocity along the particle path. The sulfate reaction process was simulated using the medium-temperature desulfurization kinetic model developed for the new calcium-based sorbent (16). The sulfate reaction rate is formulated as
dx MCaOVg ks‚kdif C ) dt m0 ks + kdif SO2
(1)
ks ) k0 exp(-Ea/RT)
(2)
kdif ) De[a exp(-bt)]
(3)
De ) De0 exp(-Ed/RT)
(4)
rdes )
where a and b are constants, x is the calcium conversion rate, MCaO is the CaO molecular weight, Vg is the gas volumetric flow rate, m0 is the CaO mass flow rate, ks and kdif are the surface reaction rate constant and product layer diffusion rate constant, respectively. CSO2 is the SO2 molar concentration. Ea and Ed are activation energies of surface reaction and production layer diffusion, respectively. De and De0 are effective and initial effective diffusion coefficients, respectively. In this model, the sulfate reaction is a first-order chemical reaction with respect to SO2 concentration with the surface reaction rate constant defined according to Arrhenius’ law. The product layer diffusion rate constant due to the diffusion near the surface has the same dimensions as the surface reaction rate constant. The effective diffusion coefficient of the gaseous reactants through the solid product layer was assumed to vary with temperature according to Arrhenius’ law. The energy equation was not solved because the reactor was isothermal. The effect of temperature on the sulfate reaction is included in the surface reaction and product layer diffusion rate constants (16). The surface reaction rate 4012
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 12, 2006
FIGURE 3. Variation of the desulfurization efficiency with time after the fresh sorbent supply was stopped at time of 120 s. constant in the reactor does not change because it is only a function of temperature and the surface reaction activation energy. However, the product layer thickness on the particle increases with increasing sorbent residence time so the gas diffusion through the solid product layer becomes more difficult, which results in a rapid decline of the product layer diffusion rate.
4. Results and Discussion 4.1. Experimental Results. The variation of the desulfurization efficiency with time after the fresh sorbent supply was stopped in the CFB facility is shown in Figure 3 for the following experimental conditions: the calcium to sulfur molar ratio, Ca/S ) 2.0; the bed superficial velocity, U ) 2.5 m/s; the bed temperature, T ) 800 °C; the bed density, Fbed ) 33.3 kg/m3; inlet SO2 concentration, 1500 ppm; inlet CO2 concentration, 10%. Figure 3 shows that, after the fresh sorbent supply stopped, the desulfurization efficiency decreased rapidly from 70% to 25% in 6 min with the maximum decline rate in the initial 60 s. Therefore, the fresh sorbent was the main contributor to the desulfurization efficiency. The primary effect of the recirculated sorbent delivered from the outer gas-solid separators was to evenly distribute the fresh sorbent and prolong the recirculation sorbent particle residence time. The product layer formed on the recirculating sorbent surface significantly reduces the product layer diffusion rate so the prolonged particle residence time resulting from the sorbent recirculation does not contribute much to the desulfurization efficiency. The increased mass flow due to the recirculated sorbent increases the bed solids concentration and enhances the particle diffusion, especially for the fresh sorbent particles which improve the desulfurization efficiency. However, high flow resistance caused by the increased recirculated sorbent mass flow is not good for engineering applications. It is very important to further investigate the effective methods to reduce the flow resistance and maintain the good diffusion for the fresh sorbent particles. 4.2. Numerical Results. Various fresh sorbent distribution types and internal structures were studied numerically to investigate their effect on the solids concentration distribution, SO2 concentration distribution, and the reactor desulfurization efficiency. The fresh sorbent distribution types were realized by the different fresh sorbent bottom injection methods. The bottom-even injection method simulates even fresh sorbent distribution in the reactor. The bottom twosided injection method simulates the fresh sorbent particles entering near both sides of the reactor. The effect of the
TABLE 1. Flow Parameters for the Fresh Sorbent Injection Methods with Various Internal Structures no.
injection method
internal no.
case no.
internal structa
1 2 3
bottom even bottom two-sided
without internals without internals single internal
case 1 case 2 case 3 case 4 case 5 case 6
without internals without internals single internal at Y ) 1.5 m single internal at Y ) 0.5 m two opposing internals at Y ) 0.5 m two alternating internals at Y ) 0.5 and 1.5 m
4 a
two internals
Y is the height of the internal structure center.
TABLE 2. Gas-Solid Flow Conditions in the Computational Model params
unit
value
SO2 mass fraction gas velocity solids velocity solids diameter solids bulk density sorbent mass CaO mass in sorbent solids concn
% m·s-1 m·s-1 mm kg·m-3 kg·s-1 kg·s-1 kg·m-3
0.5 2.5 2 83 3320 0.0256 0.0082 0.0342-0.171
number, layout, and positions of the ring-type internal structures on the gas-solid flow and the sulfate reaction characteristics was studied for the bottom two-sided injection method. The bottom two-sided injection method makes a particle flow that is the best approximation of the real annularcore particle flow structure in the CFB reactor (6). The flow parameters for the various cases are listed in Table 1. The computational domain was a two-dimensional reactor, 6 m high and 0.3 m wide. The fresh sorbent injection inlet position was at the reactor bottom. The injection inlet was equal to the reactor width for the bottom-even injection method. For the bottom two-sided injection method, the injection inlet was 0.03 m wide close to the reactor wall. The internal is triangle shape with the projected length of 60 mm. The upper internal angle with the reactor wall is 45°, and the lower internal angle is 30°. The computational conditions for the gas-solid flow are listed in Table 2. 4.2.1. Solids Concentration Distribution. The solids concentration distributions in the reactor for the various fresh sorbent injection methods with various internal structures are shown in Figure 6 for the various internal structures. Figure 4 shows that the solids concentration distribution in the bottom-even injection case 1 was uniform. For the two-sided injection case 2 without the internal structure, the particles throughout the reactor were in an annular flow pattern with little radial diffusion of the particles. Therefore, the solids concentration distribution was very uneven with high solids concentrations near the wall. Figure 4 also shows that the single internal structure disrupted the particle annular flow and improved the local solids concentration distribution near the side wall having the internal structure. The internal structure position was lower in case 4 and the influence region was longer, resulting in a better solids concentration distribution uniformity in the region above the internal structure than in case 3. Therefore, the lower height results in a better solids concentration distribution uniformity. For the designs having two internals, cases 5 and 6, the particle annular flows on both sides of the reactor wall were disrupted, further improving the solids concentration distribution uniformity. For the opposing internals, case 5, the reduced flow area resulted in a solids concentration distribution that was better than case 6. The internal structures had a significant effect on the gas and particle motion by disrupting the flow direction. Since
the particle inertia was much larger than the gas inertia, the particles moved much more radially. The particles were deflected toward the center which significantly improved the solids concentration distribution uniformity. The gas flow quickly returned to its original flow distribution with the drag force between the gas and particles being predominately in the vertical direction so that the particle radial diffusion gradually declined. The gas-solids two-phase flow will eventually come to a steady flow with minimal energy consumption, which makes the particles tend to move toward the relatively low-energy position. The gas velocity in the center was higher than that in the near-wall region, which gradually pushed the particles toward the near-wall region. The particle velocity gradually declined and the particle movement energy decreased, which induced the particles to return to particle annular flow. Therefore, the single internal only disrupts the particle annular flow and improves the local solids concentration distribution in a limited region above the internal. Above that region, the particles will gradually move back toward the near-wall region in the particle annular flow regime. The internal then divides the reactor into two particle annular flow regions, above and below the internal, with a transition region with a relatively uniform solids concentration distribution for a short distance above the internal. The distance mainly depends on the particle velocity and internal structure characteristics such as the internal angle and the projected length. To further improve the solids concentration distribution in the reactor, several internals can be placed at various heights to repeatedly produce transition regions with relatively uniform solids concentration distributions so as to reduce the particle annular flow regions. 4.2.2. SO2 Concentration Distribution. The SO2 concentration distributions in the reactor for the various fresh sorbent injection methods with various internal structures are shown in Figure 5. Figure 5 shows that, in the lower reactor region, the initially high SO2 concentration decreases rapidly, providing most of the sulfate reactions. In the upper reactor region, the SO2 concentration decreases slowly so that the upper region provides little desulfurization of the gas. The desulfurization in the middle region of the reactor is also relatively slow. The SO2 concentration distribution characteristics in the above reactor regions may result from the following reasons. In the lower reactor region, the sorbent particles had a large unreacted surface area and the product layer diffusion rate constant was much larger than the surface reaction rate constant. Therefore, the reactions in this region are controlled by the chemical kinetics. The sulfate reaction rate was very high, so the initially high SO2 concentration decreases rapidly. In the upper reactor region, the sorbent particles are covered with a product layer so that there is little unreacted surface area and the product layer diffusion rate constant is much lower than the surface reaction rate constant. Therefore, the reactions in this region are diffusion controlled. The sulfate reaction rate is then very slow, and since the SO2 concentration is also low in this region, then the SO2 concentration decreases slowly. The middle region of the reactor is called VOL. 40, NO. 12, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4013
FIGURE 4. Solids concentration distributions for the various fresh sorbent injection methods with various internal structures.
FIGURE 5. SO2 concentration distributions for the various fresh sorbent injection methods with various internal structures. the transition stage, where the reaction depends on both the chemical kinetics and diffusion through the product layer. The desulfurization in the middle is also relatively slow. 4014
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 12, 2006
In the product layer diffusion-controlled stage, the prolonged total sorbent particle residence time resulting from the sorbent particle recirculation provided little im-
Acknowledgments This research was supported by the Special Funds for Major State Basic Research Projects (No. 2006CB200305).
Literature Cited
FIGURE 6. Desulfurization efficiencies for the various fresh sorbent injection methods with various internal structures. provement to the desulfurization efficiency, which is mostly dependent on the effective residence time of the sorbent particles which have a relatively high sulfate reaction rate between the SO2 and the sorbent particles. To prolong the effective residence time, the product layer must be cracked on the sorbent surface to expose fresh sorbent surface with a high sulfate reaction rate. Some methods include enhancing the turbulence in the reactor flow field to intensify collisions and abrasion between particles or using steam activation of the sorbent particles as they are recirculated back to the reactor. It can be speculated from Figures 4 and 5 how the various internal structures influenced the SO2 concentration distributions due to the different solids concentration distributions. A uniform solids concentration distribution ensures high contact efficiency between the SO2 and the sorbent particles. Therefore, the particle annular flow needs to be frequently disrupted. The desulfurization efficiencies for the various fresh sorbent injection methods with various internal structures are compared in Figure 6. The desulfurization efficiency of the bottom-even injection method in case 1 was 1.5× that of the bottom two-sided injection method in case 2. The desulfurization efficiency in case 5 is very close to that of the bottom-even injection method in case 1 with only a 4.6% difference. When considering only the particle annular flow, the optimum layout is the opposing internal structures, due to the optimal solids concentration distribution, ensuring good contact efficiency and concentration matching between the sorbent particles and the SO2. Figure 6 also shows that the desulfurization efficiency of the opposing internals design in case 5 was 46% higher than that of case 2 without internal structures and 38% higher than that of the single internal structure in case 3 and 17% higher than that of case 4. The desulfurization efficiency of case 5 is 12% higher than that of the alternating internal design in case 6.
(1) Hou, B.; Qi, H. Y.; You, C. F. CO2 effect on medium-temperature flue gas desulfurization. J. Tsinghua Univ. 2004, 44 (11), 15711574 (in Chinese). (2) Hou, B.; Qi, H. Y.; You, C. F.; et al. New calcium-based sorbent used for medium-temperature FGD process. J. Eng. Thermophys. 2004, 25 (1), 159-162 (in Chinese). (3) Hou, B. Study on the medium-temperature dry flue gas desulfurization process. Ph.D. Thesis, Tsinghua University, Beijing, PRC, 2004 (in Chinese). (4) Grace, J. R.; Tout, J. A theory for cluster formation in vertically conveyed suspension of intermediate density. Trans. Inst. Chem. Eng. 1979, 57, 49-54. (5) Li, J. H.; Tung, Y. K.; Kwauk, M. Energy transport and regime transition in particle-fluid two-phase flow. Fluidized Bed Technology II; Basa, P., Large, J. F., Eds.; Pergamon Press: London, 1988; pp 75-87. (6) Basu, P.; Scott, A. F. Circulating fluidized bed boilers: design and operations; Butterworth-Heinemann: Stoneham, MA, 1991. (7) Li, H. Z; Guo, M. S. Particulatization of gas-solids fluidization; Chemical Industry Press: Beijing, 2002. (8) Gan, N. J.; Jiang, D. Z.; Bai, D. R.; et al. Concentration profiles in fast fluidized bed with bluff-body. J. Chem. Eng. Chin. Univ. 1990, 4 (3), 273-277. (9) Zheng, C. G.; Tung, Y. K.; Zhang, W. N.; et al. Impact of internals on radial distribution of solids in a circulating fluidized bed. Eng. Chem. Metall. 1990, 11, 296-302 (in Chinese). (10) Zheng, C. G.; Tung, Y. G.; Xia, Y. S.; et al. Internals for fast fluidized beds. In Fluidization 91, Science and Technology; Kwauk, M., Hasatani, M., Eds.; Science Press: Beijing, 1991; pp 168-177. (11) Zheng, C. G.; Tung, Y. G.; Li, H. Z.; et al. Characteristics of Fast Fluidized Beds with Internals. In Fluidization VII; Potter, O. E., Nicklin, D. J., Eds.; Engineering Foundation: New York, 1992; pp 285-293. (12) Liu, H. E.; Yang, Y. H.; Wei, F.; et al. Solids mixing behavior in riser with internals. Chem. React. Eng. Technol. 2002, 18 (2), 109-114. (13) Zhu, J. X.; Salah, M.; Zhou, Y. Radial and axial voidage distributions in circulating fluidized bed with ring-type internals. J. Chem. Eng. Jpn. 1997, 30 (5), 928-937. (14) Hou, B.; Qi, H. Y.; You, C. F.; et al. Dry Desulfurization in a Circulating Fluidized Bed (CFB) with Chain Reactions at Moderate Temperatures. Energy Fuels 2005, 19 (1), 73-78. (15) Crowe, C.; Sommerfeld, M.; Tsuji, Y. Multiphase Flows with Droplets and Particles; CRC Press: Boca Raton, FL, 1998. (16) Li, Y. R.; Qi, H. Y.; You, C. F. Kinetic model on the mediumtemperatures desulphurization of new calcium-based sorbent. Proceedings of the National Symposium on Combustion of the Chinese Society of Engineering Thermophysics; Chinese Society of Engineering Thermophysics: Beijing, 2005; pp 1093-1099 (in Chinese).
Received for review March 21, 2006. Revised manuscript received April 13, 2006. Accepted April 14, 2006. ES060665M
VOL. 40, NO. 12, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4015
