Energy Fuels 2009, 23, 6146–6151 Published on Web 10/28/2009
: DOI:10.1021/ef900713y
Modeling of Selective Catalytic Reduction (SCR) for NO Removal Using Monolithic Honeycomb Catalyst Zhigang Lei,* Xueyi Liu, and Meiru Jia State Key Laboratory of Chemical Resource Engineering, Beijing University of Chemical Technology, Box 35, Beijing, 100029, China Received July 10, 2009. Revised Manuscript Received October 8, 2009
In this study, a monolithic honeycomb catalyst is being used in a SCR reactor rather than pellet catalyst in many conventional gas-solid catalytic processes because the advantages of honeycomb matrix are very low pressure drop, high geometric specific surface area, and resistance to deposition of carbon and dust. The three-dimensional computational fluid dynamics (CFD) simulation for SCR for NO removal in monolithic honeycomb reactor was performed. The mathematical model of monolithic SCR reactor, in which a Rideal-type DeNOx rate equation was incorporated, was established. The calculated results were compared with the experimental data to verify the reliability. It was found that the lower gas inlet velocity, higher gas inlet temperature, and higher NH3/NO feed ratio lead to higher NO conversion. The square shape of the monolith channel has a better performance both in NO conversion and pressure drop. The appropriate space length between two catalyst layers under the investigated conditions is about 60 mm, which can result in good NO conversion and gas mixing. its high mechanical strength, high thermal stability, and low cost.8-11 In general, the inner walls of honeycomb-like cordierite channels are covered with a thin layer of washcoat (less than 0.1 mm) containing catalytically active species.12 To control NOx emissions from monolith honeycomb reactor, mathematical models have to be established to investigate the influences of operating parameters and design parameters of cogent industrial interest on NO conversion. So far, only one and two-dimensional steady-state isothermal mathematical models of monolith reactors for SCR of NO by NH3 have been proposed by Tronconi and Forzatti.13 Moreover, the hydraulic diameter Dh is taken as the characteristics (size and shape) of the monolith channel. In addition, the main reaction is considered and written as follows:14-17
1. Introduction Nitrogen oxides (NOx) are one of the major air pollutants in flue gas from power plants, which have to be removed before emitting into the atmosphere.1-3 As the regulation for the NOx emission becomes strict, much effort has been focused on the development of more efficient NOx removal technology. The selective catalytic reduction of NOx with ammonia (NH3) is the most effective and commercially proven technology to remove NOx from stationary sources. Commercial SCR process can remove 60-90% of NOx and has been proven to be effective at low temperature, which is beneficial for energy saving.4-7 However, structured catalysts and reactors are the key technology of SCR process. There are various types of structured catalysts, including monoliths, open crossflow structures, foams, catalytic membranes, and many others. Monolith honeycomb reactors are commonly selected because they contain a lot of parallel straight channels that are often of the order of 6-10 mm in diameter, and provide low pressure drop, high geometric specific surface area, and resistance to deposition of carbon and dust from combustion process. Besides, honeycomb-like cordierite is desirable as a support of catalysts in monolith honeycomb reactors due to
4NO þ 4NH3 þ O2 f 4N2 þ 6H2 O
The reaction is exothermic; therefore, the temperature fluctuation should exist along the monolith channel. So the focus of this work is to establish a rigorous steady-state mathematical model so as to accurately describe the reactor performance. 2. Model Description 2.1. The Equations for Gas and Solid Phases. Figure 1 shows a typical SCR reactor with cordierite honeycomb
*To whom correspondence should be addressed. Phone: þ86 10 64433695. E-mail:
[email protected]. (1) Tronconi, E.; Forzatti, P.; Martin, J. P. G.; Malloggi, S. Chem. Eng. Sci. 1992, 47, 2401–2406. (2) Tronconi, E. Catal. Today 1997, 34, 421–427. (3) Jeong, S. M.; Jung, S. H. K.; Yoo, S.; Kim, S. D. Ind. Eng. Chem. Res. 1999, 38, 2210–2215. (4) Ramachandran, B.; Herman, R.; Choi, G. S.; Stenger, H. G.; Lyman, C. E.; Sale, J. W. Catal. Today 2000, 55, 281. (5) Nam, I.; Choo, S. T.; Koh, D. J.; Kim, Y. G. Catal. Today 1997, 38, 181–186. (6) Richter, E.; Schmidt, H. J.; Schecker, H. G. Chem. Eng. Technol. 1990, 13, 332–340. (7) Hsu, L.; Teng, H. Appl. Catal. B: Environ 2001, 35, 21–30. (8) Su, J.; Liu, Q.; Liu, Z.; Huang, Z. Ind. Eng. Chem. Res. 2008, 47, 4295–4301. (9) Jeong, S. M.; Kim, S. D. Ind. Eng. Chem. Res. 2000, 39, 1911–1916. r 2009 American Chemical Society
ð1Þ
(10) Chen, J.; Yang, H.; Wang, N.; Ring, Z.; Dabros, T. Appl. Catal. A: Gen. 2008, 345, 1–11. (11) Lei, Z.; Yang, Y.; Li, Q.; Chen, B. Catal. Today 2009, 147S, S352–S356. (12) Tomasic, V.; Zrncevic, S.; Gomzi, Z. Catal. Today 2004, 90, 77– 83. (13) Tronconi, E.; Forzatti, P. AIChE J. 1992, 38, 201–210. (14) Willi, R.; Maciejewski, M.; Gobel, U. R.; Koppel, A.; Baiker, A. J. Catal. 1997, 166, 356–367. (15) Koebel, M.; Elsener, M. Chem. Eng. Sci. 1998, 53, 657–669. (16) Tronconi, E.; Cavanna, A.; Forzatti, P. Ind. Eng. Chem. Res. 1998, 37, 2341–2349. (17) Grossale, A.; Nova, I.; Tronconi, E.; Chatterjee, D.; Weibel, M. J. Catal. 2008, 256, 312–322.
6146
pubs.acs.org/EF
Energy Fuels 2009, 23, 6146–6151
: DOI:10.1021/ef900713y
Lei et al.
Figure 1. Schematic representation of SCR reactor and its square channel: (a) SCR reactor; (b) a single square channel; (c) cross-section with symmetrical boundary on the external wall and surface reaction on the inner wall.
catalysts assembled in modules, as well as its channel. The mixture of ammonia and flue gas after mixing was injected into the SCR reactor. The inner walls of monolith channel were covered with a thin layer of washcoat containing catalytically active species where SCR reaction took place. In this work, the model of a single channel with symmetrical peripheral walls was used for simulating the SCR reaction performance. The inlet fluid was the mixture of NOx (mostly NO), NH3, H2O, SO2, and air, which reacted on the inner channel walls. The flow is laminar when it enters the monolith channels. The governing equations solved by the program in this specific application are summarized as below:18 Gas Phase. Continuity equation r•ðFuÞ ¼ 0
Equation of state for ideal gas PM ¼ FRg T
The physical properties of solid phase (cordierite) are assumed to be constant. The transport properties of gas may affect the calculated results. Since a large amount of the flue gas is N2, the composition dependency can be neglected and the binary diffusion coefficients Di,N2 can be obtained as follows: Di, N2 ¼
ð2Þ
ð3Þ
Energy balance equation r•ðFcg uTÞ ¼ r•ðλg rTÞ
ð4Þ
Mass balance for species r•ðFuwi Þ ¼ r•ðFDi, N2 rwi Þ
4:36 10 -5 T 1:5 ð1=Mi þ 1=MN2 Þ0:5 1=3
PðVi
1=3
þ VN2 Þ2
ðm2 s -1 Þ ð8Þ
In the balance equations above, the physical properties of gas were taken from the process simulation software PROII at operating temperatures and pressures. To solve the governing equations, appropriate boundary conditions were specified at all external boundaries based on the following assumptions:19 (1) Uniform gas velocity, temperature and concentration at the entrance; (2) Normal pressure at the outlet; (3) Symmetrical boundary on the peripheral wall of the channel, and no slip condition on the inner wall of the channel; (4) Axially adiabatic solid boundary at the entrance and outlet. For simplification, the catalyst layer and the inner wall of the channel are taken on as overlapped due to the small thickness of the catalyst layer (less than 0.1 mm). Therefore, on the inner walls of the channel, the surface reaction as
Momentum balance equation r•ðFu uÞ ¼ -rP þ r•ðμðru þ ðruÞT Þ
ð7Þ
ð5Þ
where i = NO, NH3, O2, H2O, SO2, CO2. The carrier gas is N2. Solid Phase. Energy balance equation λs r• rT ¼ 0 ð6Þ cs (18) Hong, M.; Li, C.; Liu, H.; Ji, S. Chinese J. Chem. Eng. 2006, 14, 56–64.
(19) Roduit, B.; Wokaun, A.; Baiker, A. Ind. Eng. Chem. Res. 1998, 37, 4577–4590.
6147
Energy Fuels 2009, 23, 6146–6151
: DOI:10.1021/ef900713y
Lei et al.
boundary condition was selected, i.e., Fwall Di
Dyi, wall ¼ Mw, i Ri, gas Dn
ð9Þ
where yi,wall is the mass fraction of gas on the wall, Ri,gas is the net molar reaction rate for species i, and the effective factor of internal diffusion for global reaction rate is close to unity.20 Heat of surface reaction was also considered at wall boundary, and species diffusion effects in the energy equation due to wall surface reaction were included in the normal species diffusion term. The Gambit software (version 2.3.16) was used to mesh the monolith channel with hexagonal elements. Then, the mesh file was input into the FLUENT software (version 6.3.21) in which the laminar model was selected and the SIMPLE method was used to solve the governing equations. 2.2. Rate Equation of the DeNOx Reaction. In the mathematical model of monolith channel, the rate equation for main reaction is necessary. In line with mechanistic and kinetic evidence that either weakly adsorbed or gaseous NO reacts with NH3 strongly adsorbed on the catalyst surface, we adopted a Rideal-type rate equation over a commercial SCR catalyst as suggested by Tronconi:1,2 RNO ¼ kNO CNO
kNO ¼ 1:28 10
11
KNH3 CNH3 1 þ KNH3 CNH3
! -22400 ðs -1 Þ exp Rg T
ð10Þ
ð11Þ
In the case of R (NH3/NO feed ratio) > 1, KNH3 CNH3 . 1 and eq 10 reduces to the first order form 0
RNO ¼ kNO CNO
ð12Þ
On the contrary, with substoichiometric R (NH3/NO feed ratio) < 1, a second order form was obtained, RNO ¼ kNO KNH3 CNO CNH3
Figure 2. Influence of gas inlet velocity (area velocity AV = u d/4Z) (a) and NH3/NO feed ratio a (b) on NO conversion η: (a) side length of square channel d = 6 mm, a = 1.2, inlet temperature T = 653.15 K, z is the length of channel, 9, experimental data from ref 2 and the solid line represents the calculated results from this work. (b). square channel d = 6 mm, AV = 33 N m h-1, T = 653.15 K, 2, experimental data from ref 2, and the solid line represents the calculated results from this work.
ð13Þ
which is actually valid only in the limit of CNH3 approaching zero. 3. Results and Discussions
feed ratio) < 1, the calculated results were a little lower than the experimental data because in this case it was assumed that KNH3 CNH3 , 1 which in fact was not fair in the region near the inlet. In order to improve the accuracy of model prediction, the porous media model should be selected to account for the influence of diffusion limitation inside the porous cordierite matrix (extruded catalyst). 3.2. Influence of Operating Parameters on NO Conversion. Figure 2a and b show that the lower gas inlet velocity and higher NH3/NO feed ratio lead to higher NO conversion as we expected. The higher gas inlet velocity means the shorter residence time and thus decreased NO conversion. But NO conversion was almost unchanged when either gas inlet velocity or NH3/NO feed ratio reached at a certain value. The temperature and NO concentration profiles along monolith channel are shown in Figure 3. There was about 7 °C difference between the inlet and outlet of the channel (see Figure 3a). This indicates that the isothermal assumption made by previous researchers may be unreasonable. As shown in Figure 3b and d, the temperature on the inner wall (at X/X0 = 1.0) is the highest and herein NO concentration is
3.1. Validation of the Mathematical Model. First, the influence of grid number of the meshed geometrical volume on the calculated results was checked. Under a typical SCR operating condition, it was found that as the grid number increased, NO conversion increased rapidly at first and then became slowly. On the other hand, when the grid number N > Nmin = 1.50 106, NO conversion tends to be stable. Therefore, in our later calculation the grid number is higher than Nmin. On this basis, the calculated results were compared with the experimental data from ref 2 as shown in Figure 2. It can be seen that the calculated results in Figure 2a are a little higher than the experimental data because the surface reaction on the inner wall of monolith channel was selected in the calculation and the diffusion limitation inside the porous wall of cordierite matrix (extruded V2O5/TiO2 catalyst) was not considered. However, in Figure 2b, when R (NH3/NO (20) Avinash, R.; Sirdeshpande; JoAnn, S.; Lighty Ind. Eng. Chem. Res. 2000, 39, 1781–1787.
6148
Energy Fuels 2009, 23, 6146–6151
: DOI:10.1021/ef900713y
Lei et al.
Figure 4. Influence of shape of monolith channel on NO conversion η: O, round; 4, regular triangle; 0, square; ), rectangular; the structure and operating parameters are listed in Table 1.
Figure 5. Pressure drop ΔP for four kinds of shapes of monolith channel and the usual packed-bed reactor under the same operating condition: 4, regular triangle; O, round; ), rectangular; 0, square; b, the pressure drop in the usual packed-bed reactor with spherical particle diameter 2.5 mm; the dashed line represents the calculated results by Hagen-Poiseuille equation.
close to zero. Therefore, the DeNOx reaction takes place rapidly, and the radial mass and heat transfer is the controlled step compared with surface reaction rate. 3.3. Influence of Design Parameters on NO Conversion. The influence of shape of monolith channel, such as square, round, rectangular (the length-width ratio 3:2) and regular triangle, on NO conversion was investigated. The structure and operating parameters for these four kinds of shapes are listed in Table 1. The calculation was done under the same channel area, wall thickness and operating condition. As shown in Figure 4, at a given gas inlet area velocity, NO conversion is in the order of round > square ≈ regular triangle > rectangle. The reason may be attributed to the different gas diffusion distance for different shapes of monolith channel. Figure 5 shows the change of drop pressure between the inlet and outlet of monolith channel. It was found that at a given gas inlet velocity, the pressure drop is in the order of regular triangle > round > square > rectangle. However, in the industrial scale, the square shape is commonly selected. This can be attributed to the relatively high NO conversion, low pressure drop and resistance to deposition of carbon and dust (high hydraulic diameter) in order to ensure normal operation. Figure 5 also shows the pressure
Figure 3. Profiles of temperature along the axial central face (ZOX) of monolith channel (a), temperature at z = 0.05 m along the radial xaxis direction (b), NO concentration along the axial central face (ZOX) of monolith channel (c) and NO concentration at z = 0.05 m along the radial x-axis direction (d). The structure and operating parameters are: gas inlet temperature T = 650 K, the structure and operating parameters are: side length of square channel d = 6 mm, length of channel 717 mm, wall thickness 1 mm, gas inlet velocity u = 7 m s-1, a = 0.85 and mole fractions of inlet gas: NO 570 ppm, SO2 600 ppm, O2 0.0259, H2O 0.0738, N2 0.7451, NO2 30 ppm and CO2 0.1535.
6149
Energy Fuels 2009, 23, 6146–6151
: DOI:10.1021/ef900713y
Lei et al.
Table 1. Structure Parameters for Four Kinds of Shapes of Cordierite-Based Monolith Channela parameters
round
rectangular
regular triangle
square
49 6 6 0.5 700
49 5.812 4.716 7.573 0.5 700
49 4.987 8.906 0.5 700
49 6 6 0.5 700
2
unit area of channel, mm hydraulic diameter, mm side length of empty channel, mm wall thickness, mm length of channel, mm a
The operating conditions are inlet temperature, 650 K; NH3/NO feed ratio, 0.6; inlet mass fraction of NO, 0.000657; inlet velocity, 6 m s-1.
Table 2. Structure Parameters and NH3/NO Feed Ratio for Three Adjacent Cordierite-Based Square Channels in the First and Second Catalyst Layers. a unit area of channel, mm2 hydraulic diameter, mm wall thickness, mm length of channel, mm NH3/NO feed ratio
channel 1
channel 2
channel 3
49 6 0.5 700 0.45
49 6 0.5 700 0.50
49 6 0.5 700 0.55
a The operating conditions are inlet temperature, 650 K; inlet mass fraction of NO, 0.000657; inlet velocity, 5 m s-1.
drop of flue gas in the usual packed-bed reactor with spherical catalyst particle of 2.5 mm under the same operating condition. Evidently, the pressure drop in the usual packed-bed reactor is greater than in the monolith honeycomb reactor by 2-3 orders of magnitude. As we know, the famous Hagen-Poiseuille equation can predict the pressure drop of laminar flow through a duct. For the channel of hydraulic diameter Dh = 6 mm, the calculated results by this equation are also given in Figure 5. It can be seen that, for round and square channels, the calculated results by Hagen-Poiseuille equation are consistent with those by the mathematical model established in this work. Therefore, it goes a further step to prove that the calculated results obtained in this work are reliable. The influence of length of space between two catalyst layers (see Figure 1a), which is an important design parameter of cogent industrial interest, on NO conversion and gas mixing was investigated. The function of this space is used for remixing after DeNOx in the first catalyst layer in order to prevent from maldistribution of gas concentration into the second catalyst layer. However, the flow pattern will change abruptly from the laminar region of monolith channel in the first catalyst layer, to the turbulent region in the mixing space, and again to the laminar region of monolith channel in the second catalyst layer. Therefore, in the mixing space, the standard k-epsilon turbulence model was selected and no surface reaction occurred in our mathematical model. It was assumed that there were three adjacent square channels with different NH3/NO feed ratio, and the simulation conditions are listed in Table 2. The calculated results are shown in Figure 6. The mean variance S2 is defined as S 2 ¼ ½ðx1 -xÞ2 þ ðx2 -xÞ2 þ ðx3 -xÞ2 =3
Figure 6. Influence of the length of space between two catalyst layers L on gas mixing (a) and NO conversion (b): (a), (, calculated results, -, smoothed line; (b), (,channel 1; 9, channel 2; 2, channel 3; the structure and operating parameters are listed in Table 2.
favorable for intensifying gas mixing and controlling NH3 slip especially for some channels with high NH3/NO feed ratio, but will add equipment investment. While it reached a certain point, the changes of mean variance S2 and NO conversion were not apparent. As a result, in this case the appropriate space length is about 60 mm. 4. Conclusions The performance of monolithic SCR reactor for NO removal with NH3 was investigated by a three-dimensional computational fluid dynamics model. The calculated results are proven to be fair when compared with the experimental data. It was found that the lower gas inlet velocity, higher gas inlet temperature and higher NH3/NO feed ratio lead to higher NO conversion. But their upper and lower values are limited by production capacity of flue gas, temperature of coal power boiler upstream and maximum level of NH3 slip (below 5 ppm), respectively. The influence of design parameters, i.e., shape of monolith channel and length of space between two catalyst layers on NO conversion were also considered. It was concluded that the pressure drop in the monolith honeycomb reactor is lower than in the usual packed-bed reactor by 2-3 orders of magnitude. Besides, it should be paid more attention to determine the length of space between two catalyst layers in the design of SCR reactor since it could influence gas mixing and NO conversion.
ð14Þ
where x, x1, x2, and x3 represent the mass fraction of NO at the outlet of the second catalyst layer on the area average of three channels, on the area average of channel 1, on the area average of channel 2, and on the area average of channel 3, respectively. As the length of space between two catalyst layers increased, the mean variance S2 decreased (see Figure 6a), and uniformity of NO conversion at different channels was found (see Figure 6b). Thus, a long space is 6150
Energy Fuels 2009, 23, 6146–6151
: DOI:10.1021/ef900713y
Lei et al.
Ri,gas = net molar reaction rate for species i, mol 3 m-2 3 s-1 Rg = gas constant, J 3 mol-1 3 K-1 S2 = mean variance T = temperature, K u = gas velocity, m 3 s-1 Vi = molecular diffusion volume, m3 3 mol-1 w = mass fraction x = mass fraction of NO at the outlet of the second catalyst layer on the area average of the channel X = distance along the radial x-axis direction, m X0 = half of side length of monolith channel (= d/2), m yi,wall = mass fraction of gas on the wall z = axial distance in SCR reactor, m
Acknowledgment. This work is financially supported by the National Nature Science Foundation of China under Grant (Nos. 20736001 and 20821004), the Program for New Century Excellent Talents in University, and Fok Ying Tong Education Foundation (No. 111074).
Nomenclature AV = area velocity (= u d/4z), m 3 s-1 C = mole concentration, mol 3 m-3 C0 = mole concentration at central point, mol 3 m-3 cg = constant pressure heat capacity of gas, J 3 kg-1 3 K-1 cs = constant pressure heat capacity of solid, J 3 kg-1 3 K-1 D = gas phase diffusion coefficient, m2 3 s-1 d = channel diameter, mm Dh = hydraulic diameter, mm K = pre-exponential factor, m 3 s-1 k = intrinsic kinetic rate constant, s-1 L = length of space between two catalyst layers, mm M = molecular mass, kg 3 mol-1 N = grid number P = pressure, Pa R = rate of surface reaction, mol 3 m-2 3 s-1
Greek Symbols R = NH3/NO feed ratio η = NO conversion F = gas density, kg 3 m-3 Fwall = gas density on the wall, kg 3 m-3 μ = viscosity, Pa 3 s λg = thermal conductivity of the gas, W 3 m-1 3 K-1 λs = thermal conductivity of the solid, W 3 m-1 3 K-1
6151
