ARTICLE pubs.acs.

org/IECR

Kinetic Parameter Estimation of a Commercial Fe-Zeolite SCR

Tae Joong Wang,*,, Seung Wook Baek, Hyuk Jae Kwon,,3 Young Jin Kim, In-Sik Nam,*, Moon-Soon Cha,^ and Gwon Koo Yeo||

Propulsion and Combustion Laboratory, School of Mechanical, Aerospace and Systems Engineering, Korea Advanced Institute of Science and Technology (KAIST), 373-1 Guseong-dong, Yuseong-gu, Daejeon 305-701, Korea Environmental Catalysis Laboratory, Department of Chemical Engineering, Pohang University of Science and Technology (POSTECH), San 31, Hyoja-Dong, Pohang 790-784, Korea ^ Technology Center, ORDEG Corporation, 404 Mognae-dong, Danwon-gu, Ansan-si, Gyeonggi-do 425-100, Korea Emission Research Team, Hyundai-Kia Motors, 772-1 Jangduk-dong, Hwaseong-si, Gyeonggi-do 445-706, Korea ABSTRACT: In this work, an in-house computational code capable of simulating highly coupled physicochemical phenomena occurring in ammonia/urea SCR (selective catalytic reduction) was developed. On the basis of this computational code, the kinetic parameters of catalytic reactions were newly calibrated using the experimental results obtained over a commercial ammonia/urea SCR washcoated Fe-ion-exchanged zeolite-based catalyst. Powder-phase NH3 TPD (temperature-programmed desorption) experiments were performed to calibrate the kinetic parameters of NH3 adsorption and desorption, and core-out monolith experiments were conducted to estimate the kinetic parameters of various deNOx reactions as well as NH3 oxidation. The currently established SCR model and kinetic parameters gave a good prediction for both steady-state and transient experimental results for a wide range of operating conditions. The main objectives of this study were to develop numerical tools and their implementation methodologies that can be cost-eectively applied to the design and development of real-world ammonia/urea SCR systems. Details of the procedures and techniques in numerical modeling and kinetic parameter calibration are described step-by-step in this article.

1. INTRODUCTION Ammonia/urea selective catalytic reduction (SCR) removes NOx emissions through catalytic reactions using ammonia or urea as a reducing agent. This SCR technology has been well proven in a number of industrial stationary applications since the 1970s.1 Although there are still several problems that need to be resolved or improved, ammonia/urea SCR is thought to be the most promising technology capable of lowering diesel NOx emissions to levels required by increasingly stringent emission regulations over the world. For mobile SCR applications, aqueous urea is utilized in practice as a reductant because of the toxicity and safety problems involved in handling or transporting pure ammonia. In urea SCR, ammonia is generated in situ through thermal decomposition of urea and participates predominantly in deNOx reactions.2,3 However, there have been several reports that direct removal of NO by urea itself and its decomposition byproduct, isocyanic acid, might play an important role in overall deNOx processes.4,5 Compared with other deNOx aftertreatments such as hydrocarbon SCR or LNT (lean NOx trap), the advantages of urea SCR include higher conversion eciencies over a wider temperature window, reduced fuel penalty, greater durability, and cost savings due to the lack of precious metals. On the contrary, potential limitations of urea SCR include system complexity, costs associated with urea dosing, the absence of an infrastructure for urea, and ammonia slip.6 Selective catalytic reduction of NOx with ammonia was rst discovered over a platinum catalyst. However, platinum technology can be used only at temperatures below 250 C because of its
r 2011 American Chemical Society

poor selectivity for NOx reduction at higher temperature.1 In recent years, three major catalysts have been widely employed for urea SCR: vanadium, Cu-zeolite, and Fe-zeolite. Surely, each one shows dierent performance characteristics in view of various aspects such as operating temperature window, conversion level, human health eects, supply costs, and so on. Vanadium is relatively cheaper and more resistant to sulfur poisoning,7 but it is easily deactivated when exposed to the high temperatures required for active regeneration of soot with oxygen in diesel particulate lters (DPFs).8 It is known that limited temperature for the use of vanadium-based SCR catalysts ranges from about 600 to 650 C,9,10 whereas some V2O5/TiO2 catalysts are reported to be thermally stable up to 700 C.11 Transition-metal-promoted zeolites are able to endure high temperatures and achieve high NOx conversions. Two commonly available zeolite SCR catalysts are based on iron and copper.10 Cu-zeolite catalysts exhibit ecient NOx conversions at relatively low temperatures with little or no NO2, but display poor NOx conversions at elevated temperatures. On the other hand, Fe-zeolite catalysts show better NOx conversions at temperatures as high as 600 C or higher. However, they are not as ecient as Cu-zeolite at lower temperatures in the absence of NO2. Currently, Cu-zeolite formulations are favored when the exhaust gas temperature is lower than 450 C during the majority
Received: July 21, 2010 Accepted: December 17, 2010 Revised: November 17, 2010 Published: January 26, 2011
2850
dx.doi.org/10.1021/ie101558d | Ind. Eng. Chem. Res. 2011, 50, 28502864

Industrial & Engineering Chemistry Research of operation, whereas Fe-zeolite is preferred when the required temperature for NOx conversion exceeds 450 C.12 It is reported that the absolute upper limit on temperature for Fe-zeolite is 925 C, whereas that for Cu-zeolite is 775 C.8 Note that the latest state-of-the-art Cu-zeolite SCR catalyst shows a remarkable hightemperature hydrothermal stability up to 950 C while maintaining stable low-temperature activity in NOx conversions.13 For the past several years, modeling and simulation have been extensively employed for the design and development of mobile SCR systems. Numerical techniques are very helpful in establishing control technology of urea dosing, as well as in sustaining synchronized operation with vehicles or other aftertreatment devices such as diesel oxidation catalysts (DOCs) and DPFs. A number of studies have progressively focused on developing accurate numerical tools and their cost-eective implementation. Reaction kinetics over a catalyst is one of the most signicant factors inuencing the accuracy of mathematical models of catalytic reactor systems. However, because catalytic reactions are aected by numerous chemicophysical factors, chemical kinetics has a case-by-case nature depending on individual system conguration. This implies that the direct reuse of kinetic parameters taken from other sources cannot be justied and, therefore, that calibration is required by all means. Despite the importance of kinetic parameter calibration, it is hard to nd a fundamental work that addresses calibration processes for ammonia/urea SCR in detail. Therefore, this study is primarily intended to provide a detailed procedure and methodology for tuning kinetic parameters of various catalytic reactions occurring in ammonia/urea SCR. In this study, selected model reactions and species mass transport equations were rst mathematically described. Then, the partial dierential forms of governing equations were numerically solved using an in-house Fortran 90 computational code that was developed through this work. With this numerical tool, kinetic parameters of ammonia adsorption/desorption were newly calibrated on the basis of a powder-phase TPD experiment over a commercial Fe-zeolite catalyst. Also, kinetic parameters of various deNOx reactions and ammonia oxidation were newly estimated on the basis of core-out monolith SCR experiments performed at steady-state conditions. Finally, the simulation results produced using the current model and kinetic parameters were validated with both steady-state and transient experimental results.

ARTICLE

Table 1. Operating Conditions for NH3 TPD Analysis

parameter total experimental duration NH3 injection time NH3 shut-o time beginning time of heating for TPD temperature limit for TPD temperature increase rate for TPD initial catalyst temperature volumetric ow rate of feed gas NH3 inlet concentration value and units 12702 s 654 s 5478 s 9534 s 800 C 10 C/min 250 C 50 cm3/min 500 ppm

2. EXPERIMENTAL SECTION
2.1. NH3 TPD Experiments. An NH3 TPD analysis was carried out through a powder-phase microreactor experiment over a commercial Fe-ion-exchanged zeolite-based catalyst. For the catalyst preparation, the monolith form of a commercial SCR at 400/6.5 [cell density (cells/in.2)/wall thickness (m in.)] was crushed and ground, and then a 0.1-g sample of catalyst powder was obtained. Note that this catalyst sample contained cordierite substrate as well as catalyst itself. After the powder-phase catalyst sample was charged into a quartz tube microreactor, its pretreatment was conducted in situ at 500 C for 2 h with flowing Ar gas. The diameter of the microreactor was 10.1 mm, and the thickness of the catalyst sample layer was 2 mm. In this experiment, the reactor temperature was electrically controlled. During the whole TPD test period, the volumetric ow rate of Ar feed gas was regulated to 50 cm3/min. At 654 s, a step input of 500 ppm NH3 was admitted into the reactor, so that the inlet

concentration of NH3 increased sharply from 0 to 500 ppm. From 654 to 5478 s, a 500 ppm NH3 feed was continuously supplied at a constant reactor temperature of 250 C. This NH3 supply was shut down at 5478 s, so that the NH3 inlet concentration suddenly dropped from 500 to 0 ppm. From 5478 to 9534 s, the reactor was ushed with Ar gas at 250 C to remove the physisorbed species. From 9534 to 12702 s, the TPD experiment was performed from 250 to 800 C at a heating rate of 10 C/min with continuous monitoring of the desorbed species including NH3 (m/e = 17) and its fragment, NH2 (m/e = 16) by online mass spectrometer (Peer/Balzers Quadstar, QMI422, QME125). Operating conditions for this NH3 TPD analysis are summarized in Table 1. 2.2. Steady-State SCR Experiments. The catalytic activities of the same commercial SCR catalyst as used in the NH3 TPD experiment were examined using an integral flow reactor system under steady-state conditions as shown in Figure 1. For these tests, a core part was taken from the original SCR catalyst so that a small (diameter length = 20 mm 40 mm) core-out monolith SCR catalyst was prepared. The following list provides the feed gas compositions (commonly containing 5% O2, 10% H2O, and balance N2) and reactor space velocities for each reaction test: (1) NH3 oxidation, 500 ppm NH3 at 10000 h-1; (2) NO SCR reaction, 500 ppm NH3 and 500 ppm NO at 10000 and 15000 h-1; (3) NO2 SCR reaction, 500 ppm NH3 and 500 ppm NO2 at 30000-50000 h-1; and (4) NOx SCR reaction, 500 ppm NH3, 250 ppm NO, and 250 ppm NO2 at 3000050000 h-1. Note that reactor space velocity is defined as the ratio of the total volumetric flow rate to the volume occupied by the monolith reactor. In these core-out monolith SCR experiments, all measurement data were obtained after pretreatment of the catalysts under air atmosphere at 500 C for 2 h. Concentrations of NO, NO2, and NH3 were measured by online chemiluminescence NO-NOx analyzer (Thermo Electron Corporation, model 42H), NO2 analyzer equipped with an electrochemical cell (Testo, model 350M), and a nondispersive-infrared(NDIR-) type NH3 analyzer (Rosemount Analytical, model 880A), respectively. 2.3. Transient SCR Experiments. The transient catalytic activities of the same size of core-out monolith SCR as used in the previous steady-state experiments were also evaluated using the reactor system illustrated in Figure 1. In these transient tests, the inlet concentrations of NO and NO2 were changed with time at constant reactor temperatures of 230 and 300 C. Here, the NH3 inlet concentration was constantly set to 600 ppm. The sum of the NO and NO2 inlet concentrations was also regulated to 600 ppm, while their ratio was changed with time as follows:
2851
dx.doi.org/10.1021/ie101558d |Ind. Eng. Chem. Res. 2011, 50, 28502864

Industrial & Engineering Chemistry Research

ARTICLE

Figure 1. Schematic ow diagram of the NH3 SCR reactor system.

NO/NO2 = (1) 600 ppm/0 ppm f (2) 500 ppm/100 ppm f (3) 400 ppm/200 ppm f (4) 300 ppm/300 ppm f (5) 200 ppm/400 ppm f (6) 100 ppm/500 ppm f (7) 0 ppm/600 ppm. The uniform feed of NO and NO2 at each step was maintained for 3 min except for step 1, which was continued for 13 min because the transient test was started with the monolith SCR being fresh and, therefore, sufficient time was required for the catalyst surface to reach its steady state. Also, the time interval between each step from 1 to 7 was regulated as follows: step 1 f 2 = 3 min, step 2 f 3 = 2 min, step 3 f 4 = 1 min, step 4 f 5 = 0.5 min, step 5 f 6 = 1 min, step 6 f 7 = 2 min. For all test periods, the feed gas composition was set to contain 5% O2, 10% H2O, and balance N2, and the reactor space velocity was uniformly maintained at 40000 h-1.

which gaseous NH3 molecules are adsorbed or desorbed with 1:1 adsorption stoichiometry. This assumption has been widely adopted in the literature for model simplicity.17-20 The adsorption and desorption of NH3 on an active site are described by the following forward and backward reactions NH3 S T NH3 3 S 1

Here, the symbol S denotes an active surface site, and the NH3 adsorption and desorption rates are modeled as Ra ka CNH3 1 - NH3 Rd kd NH3 2 3

3. MODELING
3.1. Reaction Kinetics and Mass Balances. NH3 Adsorption/ Desorption. In ammonia/urea SCR, NH3 molecules are chemi-

respectively, where each rate constant is formulated by an Arrhenius form such that   Ei o ki ki exp , i a, d 4 Ru Ts To determine the adsorption/desorption activation energies in eq 4, we have referred to other works conducted over vanadium-based catalysts.21,22 Therefore, in this model, NH3 adsorption is assumed to occur through a nonactivated process so that the adsorption activation energy, Ea, is set to zero. On the other hand, the NH3 desorption process is assumed to be strongly dependent on surface conditions so that the desorption activation energy, Ed is represented by a Tempkin-type expression in which Ed decreases linearly with increasing NH3 surface coverage, NH3, as Ed Eo 1 - RNH3 d
2852

cally adsorbed both on active metal sites and on Brnsted o acid sites.14 Because NH3 plays a primary role in deNOx processes, modeling of NH3 adsorption/desorption over the catalyst surface is an essential part of the overall SCR model. Hence, NH3 adsorption/desorption should be realistically modeled for an accurate prediction of the performance of ammonia/urea SCR. It is known that several NH3 molecules can be bound to one active metal site. Komatsu et al.15 reported that each copper ion can coordinate up to four NH3 molecules. Recently, based on this report, Olsson et al.16 presented an SCR model with detailed surface descriptions. However, the current model assumes that there exists only one kind of surface site (denoted by S in eq 1) on

dx.doi.org/10.1021/ie101558d |Ind. Eng. Chem. Res. 2011, 50, 28502864

Industrial & Engineering Chemistry Research The mass balance of NH3 in the gas phase is described as p DCg, NH3 DCg, NH3 - uD, p ac, p Rd - Ra Dt Dx 6

ARTICLE

where the left-hand side indicates the unsteady accumulation of gas-phase NH3 within the overall packed-bed reactor-averaged control volume, which consists of both pore and solid parts. On the right-hand side, the rst term physically represents the convective ux of gas-phase NH3. Note that the supercial velocity (often referred to as the Darcian velocity), uD,p is equal to the pore velocity (i.e., the actual velocity through channel) multiplied by the void fraction of the packed-bed reactor. Also, the second term accounts for the source or sink of gas-phase NH3 through its adsorption or desorption on active sites. There must be a careful identication in dening catalytic active site because it is diversely represented in the literature.19,22 The mass balance of NH3 on the solid surface can be written as DNH3 Ra - Rd Dt 7

Figure 2. Measured conversion of NH3 versus catalyst temperature for steady-state NH3 oxidation. Feed gas composition: 500 ppm NH3, 5% O2, 10% H2O, and balance N2. Space velocity = 10000 h-1.

The model reaction of NH3 oxidation is given by 4NH3 3 S 3O2 f 2N2 6H2 O 4S The rate of NH3 oxidation is expressed as Rox kox CO2 NH3 10 9

which physically describes that the dierence between the NH3 adsorption and desorption rates causes a temporal change in NH3 accumulation on the active sites. In eq 6, the void fraction of the packed bed, p, can be obtained using the relation F b, p 1 - m 8 p 1 F b, m where measurements on the masses and geometries of the monolith and packed bed give the bulk density of the monolith as Fb,m = 462.46 kg/m3 and the bulk density of the packed bed as Fb,p = 624.08 kg/m3. Also, from a simple calculation based on the monolith cell density and wall thickness, the void fraction of the monolith, m, is 0.5476. With these known properties, eq 8 nally yields p = 0.3895. NH3 Oxidation. Accurate prediction of NH3 concentration through ammonia/urea SCR is important in establishing a reliable SCR model because NH3 is related to a variety of deNOx reactions and to NH3 slip. Therefore, the current SCR model includes a model of NH3 oxidation with O2 in addition to models of various deNOx reactions with NH3. The NH3 oxidation capability of general SCR catalysts is quite weak up to relatively high temperatures, and also at low temperatures. Wurzenberger and Wanker17 reported that NH3 oxidation over SCR catalysts is not pronounced until the temperature reaches as high as 500 C. In the present steady-state NH3 oxidation experiments on a core-out monolith SCR catalyst, NH3 concentrations were measured at both the inlet and outlet of the SCR catalyst with increasing temperature at a constant space velocity of 10000 h-1. It should be noted that the current space velocity level is quite low because the activity for NH3 oxidation of this Fe-based SCR catalyst is relatively weak and, therefore, a sucient range of conversion rates required for a reliable kinetic parameter calibration cannot be obtained at higher space velocities. The experimental results are presented in Figure 2, which reveals that NH3 oxidation with O2 begins at about 350 C and its rate becomes higher as temperature increases. Also, it is observed that the conversion rate of NH3 through its oxidation reaches approximately 60% at 500 C.

where the rate constant is modeled as the following Arrhenius form   Eox o kox kox exp 11 Ru Ts To simulate the NH3 oxidation process through the reactor, the mass balances of gas- and solid-phase NH3 and gas-phase O2 should be solved simultaneously. First, the gas-phase NH3 mass balance is described by eq 6 as well because the participation of gas-phase NH3 in its oxidation with O2 is negligible; rather, NH3 oxidation occurs in the adsorbed phase as expressed in reaction 9. Second, the solid-phase NH3 mass balance is represented dierently from eq 7 because the consumption of adsorbed NH3 through its oxidation should be further taken into account; therefore, it is written as DNH3 Ra - Rd - Rox Dt 12

Third, because gas-phase O2 takes part in NH3 oxidation as described in eq 10, its concentration should be known along the reactor. This requires the following mass balance of gas-phase O2 to be solved DCg, O2 DCg, O2 3 - uD, m - ac, m Rox m 13 4 Dt Dx Here, the coecient of 3/4 multiplying Rox was obtained by normalizing reaction 9 with respect to adsorbed-phase NH3 (NH3 3 S). NO SCR Reaction. In a typical engine-out exhaust, NO is the major NOx component. It is well-known that the SCR of NO by NH3 is a dominant deNOx reaction pathway in ammonia/urea SCR. Thus, a model of the NO SCR reaction takes a critical role in establishing an accurate overall SCR model. In the current steady-state NO SCR experiments on a core-out monolith SCR catalyst, the concentrations of NO and NH3 were measured at both the inlet and outlet of the catalyst with
2853
dx.doi.org/10.1021/ie101558d |Ind. Eng. Chem. Res. 2011, 50, 28502864

Industrial & Engineering Chemistry Research

ARTICLE

This is the so-called standard SCR reaction, and its rate expression is described as 15 RNO kNO CNO NH3 where the rate constant is modeled as the following Arrhenius form   ENO kNO ko exp 16 NO Ru Ts To simulate the NO SCR process through the reactor, the mass balances of gas- and solid-phase NH3, gas-phase O2, and gas-phase NO should be solved simultaneously. First, the gas-phase NH3 mass balance is described by eq 6 because the NH3 molecule participates in various SCR reactions as its adsorbed phase, not as its gas phase. Second, the solid-phase NH3 mass balance is expressed dierently from either eq 7 or eq 12 because the consumption of adsorbed NH3 through the NO SCR reaction should be further taken into account. Therefore, it is represented by
Figure 3. Measured conversions of (a) NO and (b) NH3 versus catalyst temperature for the steady-state NO SCR reaction. Feed gas composition: 500 ppm NH3, 500 ppm NO, 5% O2, 10% H2O, and balance N2.

DNH3 Ra - Rd - Rox - RNO Dt

17

increasing temperature at constant space velocities of 10000 and 15000 h-1. In these experiments, the temperature and space velocity levels were actually selected such that the measured conversion rates varied over a broad range, thereby ensuring reliability and applicability of the newly calibrated kinetic parameters. The experimental result are shown in Figure 3, where it is observed that the conversion rates of NO and NH3 are almost the same. This indicates that the consumption ratio of NO/ NH3 is nearly 1:1, which strongly supports the NO/NH3 stoichiometry given in reaction 14. The light-o temperature of NO is about 210 C at 10000 h-1 and about 227 C at 15000 h-1, and the measured temperature at which 100% NO conversion rst appeared was 276.5 C at 10000 h-1 and 316.5 C at 15000 h-1. Complete conversion of NH3 was observed over a wide temperature range, and the NH3 conversion rate never decreased once it reached 100%. A similar behavior was also observed for NO; hence, the NO conversion rate never decreased once it reached 100%, although it is expected that NH3 oxidation occurs at temperatures over around 300 C so that the amount of NH3 required for complete NO consumption was not sucient at this high temperature region considering the NO/NH3 stoichiometry of 1:1. This can be attributed to the fact that, when the feed gas contains equal amounts of NH3 and NO with abundant O2, almost all of the NH3 is consumed through the SCR reaction with NO and O2 than through NH3 oxidation with O2 because the NO SCR reaction is much faster than NH3 oxidation. Also, from the experimental results, it can be inferred that the current Fezeolite catalyst has a good capability in adsorbing NH3 molecules because NO conversion does not decrease even at high temperatures near 500 C. For a vanadium-based catalyst, it has been reported that the NO removal activity decreases above 380 C because of its weak NH3 adsorption performance at high temperatures.23 The model reaction of NO SCR is given by 4NH3 3 S 4NO O2 f 4N2 6H2 O 4S 14

Third, the gas-phase O2 mass balance is given by   DCg, O2 DCg, O2 3 1 - uD, m - ac, m Rox RNO 18 m 4 4 Dt Dx where the coecients 3/4 multiplying Rox and 1/4 multiplying RNO were obtained by normalizing reactions 9 and 14, respectively, with respect to adsorbed-phase NH3. Fourth, the gas-phase NO mass balance is expressed as m DCg, NO DCg, NO - uD, m - ac, m RNO Dt Dx 19

where the coecient 1 multiplying RNO results from normalizing reaction 14 with respect to adsorbed-phase NH3. NO2 SCR Reaction. In general, for common ammonia/urea SCR catalysts based on vanadium or zeolite, the deNOx performance is enhanced as the NO2 concentration increases because the reaction pathway consuming NO and NO2 simultaneously is very fast. However, as the ratio of NO2 to NOx becomes too high, the overall deNOx activity of the SCR catalyst is lowered again because the reaction pathway removing NO2 only is slow. Koebel et al.24 reported that, for a vanadium-based catalyst, the NO2/ NOx ratio should not exceed 0.5 to maximize the deNOx performance, whereas Baik et al.25 reported that the optimum NO2/NOx feed ratio for the best deNOx activity is 0.75 for both Fe-ZSM5 and Cu-ZSM5 catalysts and 0.5-0.75 for V2O5/TiO2 catalyst. Also, Baik et al.25 concluded that the optimum NO2/ NOx feed ratio depends on the catalyst and its temperature. Therefore, for an accurate prediction of SCR performance at various NO/NO2 compositions and operating conditions, the SCR of NO2 by NH3 should be included in the overall SCR model. In the present steady-state NO2 SCR experiments on a coreout monolith SCR catalyst, the concentrations of NO2 and NH3 were measured at both the inlet and outlet of the catalyst with increasing temperature at constant space velocities of 30000, 40000, and 50000 h-1. The experimental result is illustrated in Figure 4, where the lowest temperature and space velocity levels were higher than those employed in the previous NO SCR experiment. This is because, as the temperature or space velocity
2854
dx.doi.org/10.1021/ie101558d |Ind. Eng. Chem. Res. 2011, 50, 28502864

Industrial & Engineering Chemistry Research

ARTICLE

To simulate the NO2 SCR process through the reactor, the mass balances of gas- and solid-phase NH3, gas-phase O2, and gas-phase NO2 should be solved simultaneously. First, the gas-phase NH3 mass balance is described by eq 6 because the reactions between all the gas-phase molecules are negligible. Second, the solid-phase NH3 mass balance should consider the consumption of adsorbed NH3 by NO2 SCR reaction and therefore it is represented as DNH3 Ra - Rd - Rox - RNO2 25 Dt Third, gas-phase O2 mass balance is also the same as eq 13 because gas-phase O2 does not participate in the NO2 SCR reaction. Fourth, on the basis of the 1:1 stoichiometry for NH3/NO2, the gas-phase NO2 mass balance is expressed as DCg, NO2 DCg, NO2 - uD, m - ac, m RNO2 26 m Dt Dx
Figure 4. Measured conversions of (a) NO2 and (b) NH3 versus catalyst temperature for the steady-state NO2 SCR reaction. Feed gas composition: 500 ppm NH3, 500 ppm NO2, 5% O2, 10% H2O, and balance N2.

becomes lower than the currently adopted level, the formation of ammonium nitrite (NH4NO2)26 through the gas-phase homogeneous reaction begins to inuence the experimental results. In Figure 4, the comparison between NO2 and NH3 conversions reveals that the reaction stoichiometry of NO2/NH3 is almost 1:1. Also, it is observed that both the NO2 and NH3 conversion rates decrease with increasing space velocity, whereas they increase with increasing temperature. Note that, unlike the previous NO conversion, the NO2 conversion is lowered again at temperatures above around 350 C. Perhaps, this is mainly due to NH3 oxidation in the corresponding temperature zone. The model reaction of the NO2 SCR reaction is 4NH3 3 S 3NO2 f 7 N2 6H2 O 4S 2 20

This reaction indicates that the stoichiometric ratio of NH3 to NO2 is 4:3. However, the current experimental result does not follow this ratio; rather, it shows a 1:1 stoichiometry. This can be attributed to the fact that reactions 21 and 22 to produce N2O occur within SCR in addition to reaction 2026 3NH3 3 S 4NO2 f 7 9 N2 O H2 O 3S 2 2 21 22

4NH3 3 S 4NO2 O2 f 4N2 O 6H2 O 4S

where the coecient 1 multiplying RNO2 results from normalizing reactions 20-22 with respect to adsorbed-phase NH3 and integrating them. NOx SCR Reaction. Rapid NOx conversion through the reaction consuming equal amounts of NO and NO2 with NH3 has long been known,27,28 which is considered as a practical means to increase the performance of ammonia/urea SCR. Therefore, the fast SCR reaction consuming equimolar NO and NO2 (referred to as the NOx SCR reaction in this study) is finally modeled here. In the current steady-state NOx SCR experiments on a coreout monolith SCR catalyst, the concentrations of NO, NO2, and NH3 were measured at both the inlet and outlet of the catalyst with increasing temperature at constant space velocities of 30000, 40000, and 50000 h-1. The experimental results are displayed in Figure 5, where a comparison of the NOx and NH3 conversions reveals that the NOx/NH3 consumption ratio is nearly 1:1. This strongly supports the (NO NO2)/NH3 stoichiometry given in reaction 27. For a more detailed examination of the measured NOx conversion results, each conversion rate of NO and NO2 is separately displayed in Figure 5c, where the NO2 conversion performance exceeds that for NO: NO2 shows 100% conversion over the entire temperature range, whereas NO does not. This indicates that the consumption rate of NO2 is higher than that of NO.29 This is because NO2 plays an important role in reoxidizing the active site on the catalyst surface reduced in the activation (N2-) of NH3. Note that NO2 is a stronger oxidant than O2.28,30 The model reaction of the NOx SCR reaction is 27 4NH3 3 S 2NO 2NO2 f 4N2 6H2 O 4S The rate expression of reaction 27 is written as RNOx kNOx CNO CNO2 NH3 28

The sum of reactions 20-22 leads to a 1:1 reaction stoichiometry for NH3/NO2. Therefore, the rate of the NO2 SCR reaction is expressed as RNO2 kNO2 CNO2 NH3 23

where the rate constant is modeled as the following Arrhenius form   ENO2 o 24 kNO2 kNO2 exp Ru Ts

where the rate constant is modeled as the following Arrhenius form   ENOx o kNOx kNOx exp 29 Ru T s To simulate the NOx SCR process through the reactor, the mass balances of gas- and solid-phase NH3 , gas-phase O2, gas-phase NO, and gas-phase NO2 should be solved simultaneously.
2855
dx.doi.org/10.1021/ie101558d |Ind. Eng. Chem. Res. 2011, 50, 28502864

Industrial & Engineering Chemistry Research

ARTICLE
1 where the coecients 1 multiplying RNO2 and /2 multiplying RNOx were obtained by normalizing reactions 20-22 and 27, respectively, with respect to adsorbed-phase NH3. Boundary and Initial Conditions. Adequate boundary and initial conditions are required to obtain a complete set of solutions for the above governing equations. Gas-phase mass balances commonly require two boundary conditions (i.e., the Dirichlet condition at inlet and the Neumann condition at outlet) and one initial condition as follows

Cg, i t, x 1 given at inlet, i NH3 , O2 , NO, NO2 DCg, i t, x Imax 0, Dx Cg, i t 0, x 0, i NH3 , O2 , NO, NO2 i NH3 , O2 , NO, NO2 33 34 35

The solid-phase NH3 mass balance requires no boundary condition but only one initial condition as follows NH3 t 0, x 0 36

3.2. Numerical Solutions. Discretization of Governing Equations. As a first step to obtaining the numerical solutions of the

Figure 5. Measured conversions of (a) NOx, (b) NH3, and (c) NO and NO2 versus catalyst temperature for the steady-state NOx SCR reaction. Feed gas composition: 500 ppm NH3, 250 ppm NO, 250 ppm NO2, 5% O2, 10% H2O, and balance N2.

First, the gas-phase NH3 mass balance is the same as in eq 6 because the reactions between all the gas-phase molecules are negligible. Second, the solid-phase NH3 mass balance is represented as DNH3 Ra - Rd - Rox - RNO - RNO2 - RNOx Dt 30

Third, the gas-phase O2 mass balance is the same as in eq 18, which is the governing equation of the NO SCR reaction because gas-phase O2 does not participate in both the NO2 and NOx SCR reactions. Fourth, the gas-phase NO mass balance is expressed as   DCg, NO DCg, NO 1 - uD, m - ac, m RNO RNOx m 2 Dt Dx 31 where the coecients 1 multiplying RNO and 1/2 multiplying RNOx were obtained by normalizing reactions 14 and 27, respectively, with respect to adsorbed-phase NH3. Fifth, the gas-phase NO2 mass balance is given by   DCg, NO2 DCg, NO2 1 - uD, m - ac, m RNO2 RNOx m 2 Dt Dx 32

governing equations that were defined above, the SCR reactor was represented by a one-dimensional computational domain divided into 200 equally spaced small cells. In this study, a cellcentered coordinate system was adopted so that the total number of nodal points was 202 (i.e., Imax = 202); two nodes coincided with both the inlet and outlet cell faces, and the other 200 nodes were placed on the center of each cell. Then, partial dierential forms of the governing mass balances were discretized on each computational node to yield their linearized forms. In this study, the gas-phase mass balances were discretized using the Euler implicit method,31 which was unconditionally stable for all time steps and second-order-accurate for space and rst-order-accurate for time, by employing central space-dierencing and forward time-dierencing schemes. The solid-phase mass balances were discretized using a simple forward time-dierencing scheme, which resulted in rst-orderaccurate solutions for time. Numerical Techniques. With the boundary and initial conditions, the linearized governing equations were numerically solved regarding the kinetics of each reaction by developing an in-house computational code written in Fortran 90. In obtaining the numerical solutions, an iterative calculation technique was implemented because the gas and solid phases are coupled each other. As a convergence criterion, the iterative calculation was continued until the error dened in eq 37 reached as small as 10-3 at every time step   Imax - 1 X  n  X Y - Y n - 1  n 37     Yn x2 Y where the superscript n indicates current iteration step and n - 1 denotes previous iteration step at the same time step. Although this error calculation formula was identical for all reactions,
2856
dx.doi.org/10.1021/ie101558d |Ind. Eng. Chem. Res. 2011, 50, 28502864

Industrial & Engineering Chemistry Research the variable Y diered for each reaction, as summarized in Table 2. To prevent solution divergence, this study employed an under-relaxation technique in which an under-relaxation factor of 0.2 was used in calculating each reaction rate except for the NOx SCR reaction (0.15 was employed for this reaction). Actually, these values were determined after many preliminary calculations based on the tradeo between speed and stability in solution convergence; a larger under-relaxation factor causes greater speed but poorer stability, and vice versa. Setting an incremental time for obtaining simulation results also required trial and error because there is a tradeo between solution accuracy and computing duration. In simulating the Table 2. Convergence Criterion Variables for Each Reaction
reaction NH3 adsorption/desorption NH3 oxidation NO SCR NO2 SCR NOx SCR convergence criterion variable Y = NH3, Cg,NH3, Ra, Rd Y = NH3, Cg,NH3, Cg,O2, Ra, Rd, Rox Y = NH3, Cg,NH3, Cg,O2, Cg,NO, Ra, Rd, Rox, RNO Y = NH3, Cg,NH3, Cg,O2, Cg,NO2, Ra, Rd, Rox, RNO2 Y = NH3, Cg,NH3, Cg,O2, Cg,NO, Cg,NO2, Ra, Rd, Rox, RNO, RNO2, RNOx

ARTICLE

NH3 TPD analysis, a time increment of 1 s was used. In obtaining the steady-state simulation results for NH3 oxidation, NO SCR, NO2 SCR, and NOx SCR, the transient calculations were continued until steady state was reached for each operating condition. Here, the time increment of transient simulation was set to 10 s. Also, in obtaining transient simulation results for the NOx SCR reaction, a time increment of 1 s was employed.

4. KINETIC PARAMETER ESTIMATION

NH3 Adsorption/Desorption. Using the currently developed computational code, kinetic parameter calibrations for NH3 adsorption and desorption were carried out on the basis of the experimental results. Figure 6 displays a comparison of the calculated and measured results on NH3 TPD analysis, which shows a fairly good agreement. Here, the simulation results were obtained using a trial-and-error method in which the kinetic parameters of NH3 adsorption and desorption were altered until the deviation between simulation and experiment was minimized. Both the initial and newly calibrated kinetic parameters are summarized in Table 3. In Figure 6, the initial stage reveals that the NH3 exit concentration is equal to the inlet concentration (500 ppm) after some duration due to the adsorption of NH3 onto the catalyst surface. A similar behavior was detected after the shut-o of NH3 feed. From about 5500 s, the NH3 exit concentration slowly decreased with time because the adsorbed NH3 desorbed

Figure 6. Measured and simulated NH3 exit concentrations together with NH3 inlet concentration and reactor temperature in NH3 TPD experiments.

Figure 7. Arrhenius plot with a least-squares regression curve to newly tted pre-exponential factors of the NH3 oxidation rate. Curve information: slope = 4459.07, intercept = 13.6364.

Table 3. Kinetic Parameters of NH3 Adsorption and Desorption

parameter pre-exponential factor of the NH3 adsorption rate, ko a pre-exponential factor of the NH3 desorption rate, ko d Eo d activation energy of the NH3 desorption rate at zero coverage, parameter for NH3 surface coverage dependence, R active-site density based on the solid part of monolith, ac,m/m initiala 0.93 m3/(mol s) 1.0 10 s 0.98 200.0 mol/m3
11 -1

newb 0.6 m3/(mol s) 2.0 1010 s-1 180.0 kJ/mol 0.7 200.9 mol/m3

181.5 kJ/mol

a Initial values adopted from Olsson et al.19 b New values obtained through the best t of the simulation to NH3 TPD experiments over a commercial Fezeolite catalyst.

2857

dx.doi.org/10.1021/ie101558d |Ind. Eng. Chem. Res. 2011, 50, 28502864

Industrial & Engineering Chemistry Research Table 4. Kinetic Parameters for NH3 Oxidation and Various deNOx Reactions
pre-exponential factor [m3/(mol s)] reaction NH3 oxidation NO SCR reaction NO2 SCR reaction
a

ARTICLE

activation energy (kJ/mol) initiala 162.4 84.9 72.3 85.1 newb 125.3 77.2 49.3 67.1

initiala 1.2 1011 2.3 10

8

newb 8.36 105 7.48 106 2.55 104 2.50 109

1.1 107 1.9 1012

Initial values adopted from Olsson et al.19 b New values obtained through the best t of simulation to steady-state SCR experiment over a commercial Fe-zeolite catalyst.

NOx SCR reaction

Figure 8. Comparison of calculated and measured NH3 exit concentrations for steady-state NH3 oxidation.

from the catalyst surface. From about 9500 s, NH3 desorption was enhanced by the heating of the catalyst (TPD experiment), and nally, complete desorption of NH3 was achieved. In Table 3, it should be noted that the value of the catalytic active-site density is given based on a monolith because the modeling of other reactions was conducted for a monolith conguration. Actually, the active site was rst obtained based on a packed bed and was then converted to a monolith-based value. The conversion was performed using the following simple relation between monolith and packed-bed properties F b, p 38 ac, p ac, m F b, m The current calibration rst gives the active-site density based on the packed-bed bulk volume, ac,p = 148.4 mol/m3. Then, eq 38 yields the active-site density based on the monolith bulk volume, ac,m = 110.0 mol/m3. In the literature, the active-site density is often dened based on only the solid part of catalyst, not the bulk.19,22 Therefore, for a convenient comparison, the active-site density in Table 3 is given based on the solid part of the monolith SCR catalyst. Note that a solid-part-basis property can be simply obtained from a bulk-basis property by dividing by the void fraction. NH3 Oxidation. For a realistic simulation of NH3 oxidation, NH3 adsorption and desorption should be considered together. This is because the oxidation of NH3 takes place in its adsorbed state onto active sites, so that the NH3 oxidizing capability is dependent on its adsorption/desorption characteristics. Therefore, in this study, kinetic parameter tuning for NH3 oxidation was performed using the already-estimated kinetics of NH3 adsorption/desorption. Using the currently developed computational code, kinetic parameter calibration for NH3 oxidation was carried out on the

Figure 9. Arrhenius plot with a least-squares regression curve to newly tted pre-exponential factors of the NO SCR reaction rate. Curve information: slope = 931.536, intercept = 15.8271.

basis of the steady-state experimental results given in Figure 2. In this calibration, only the measurement data obtained above 345.5 C (i.e., four high-temperature data points in Figure 2) were employed because other data measured below 285.5 C exhibited zero NH3 conversion. An Arrhenius plot for the pre-exponential factors of the NH3 oxidation rate that were newly tted is illustrated in Figure 7. To obtain these numerical t results, the iterative calculation was continued until the deviation of the calculated NH3 exit concentration from the measured value became as small as 0.1 ppm. The initial values of the kinetic parameters employed here were taken from Olsson et al.19 and are listed in Table 4. From the Arrhenius plot shown in Figure 7, both the pre-exponential factor and activation energy of NH3 oxidation rate were newly determined, as also summarized in Table 4. A detailed procedure to derive those new kinetic parameters can be found in Wang et al.32 In Figure 8, a comparison of the simulation results obtained using the newly calibrated kinetic parameters with the experimental results is exhibited, and the agreement between them is quite good. NO SCR Reaction. For an accurate simulation of the NO SCR reaction, NH3 adsorption/desorption and NH3 oxidation should be considered together. Therefore, in this study, kinetic parameter tuning for the NO SCR reaction was performed using the already-estimated kinetics of NH3 adsorption/desorption and NH3 oxidation.
2858
dx.doi.org/10.1021/ie101558d |Ind. Eng. Chem. Res. 2011, 50, 28502864

Industrial & Engineering Chemistry Research

ARTICLE

Figure 10. Comparisons of calculated and measured exit concentrations and conversion rates of (a) NO and (b) NH3 for the steady-state NO SCR reaction. Space velocity = 10000 h-1.

Figure 12. Arrhenius plot with a least-squares regression curve to newly tted pre-exponential factors of the NO2 SCR reaction rate. Curve information: slope = 2767.15, intercept = 10.1469.

Figure 11. Comparisons of calculated and measured exit concentrations and conversion rates of (a) NO and (b) NH3 for the steady-state NO SCR reaction. Space velocity = 15000 h-1.

Figure 13. Comparisons of calculated and measured exit concentrations of (a) NO2 and (b) NH3 for the steady-state NO2 SCR reaction. Space velocity = 30000 h-1.

On the basis of the currently developed numerical tools, the kinetic parameters of the NO SCR reaction were calibrated using the steady-state experimental results given in Figure 3. In this calibration, both the data sets measured at 10000 and 15000 h-1 were simultaneously utilized. However, several high-temperature data were excluded because of their complete NO conversion. The exact magnitude of the reaction rate cannot be estimated from the data with zero exit concentration because reaction rates larger than a certain threshold value, corresponding to complete NO consumption exactly at the SCR exit, also yield zero NO emissions. In addition, some measurement data located far from the regression curve on the Arrhenius plot were also kept out of the calibration. In the end, four data points at 10000 h-1 and three data points at 15000 h-1 were actually employed for this calibration.

Figure 9 presents an Arrhenius plot for the pre-exponential factors of the NO SCR reaction rate that were newly tted. To obtain each data point in Figure 9, an iterative calculation was performed while updating the pre-exponential factor. The stopping criterion was that the deviation between the calculated and measured species concentrations at the SCR exit became as small as 0.1 ppm. In this study, because NO and NH3 were both employed for stopping criteria, two tted data sets were produced at one catalyst temperature. Analyzing the Arrhenius plot with reference to Wang et al.32 gave a new pre-exponential factor and activation energy of the NO SCR reaction rate, as listed in Table 4. The initial values of the kinetic parameters used in this calibration were taken from Olsson et al.19 and are also summarized in Table 4. Figures 10 and 11 compare the calculated and measured results for the NO SCR reaction at space velocities of 10000
2859
dx.doi.org/10.1021/ie101558d |Ind. Eng. Chem. Res. 2011, 50, 28502864

Industrial & Engineering Chemistry Research

ARTICLE

Figure 14. Comparisons of calculated and measured exit concentrations of (a) NO2 and (b) NH3 for the steady-state NO2 SCR reaction. Space velocity = 40000 h-1.

Figure 16. Arrhenius plot with a least-squares regression curve to newly tted pre-exponential factors of the NOx SCR reaction rate. Curve information: slope = 2165.30, intercept = 21.6393.

Figure 15. Comparisons of calculated and measured exit concentrations of (a) NO2 and (b) NH3 for the steady-state NO2 SCR reaction. Space velocity = 50000 h-1.

and 15000 h-1, respectively. A good agreement is observed for both species and for both space velocities. The simulation results show that the conversion rate of NO reaches 100% above around 300 C, and then, it becomes slightly lower with further increase in temperature, which was not observed in the experiments. This dierence can be attributed to the fact that the rate of NH3 oxidation with O2 becomes appreciable in the corresponding high-temperature region and, therefore, the amount of NH3 required to react with NO is not sucient there. NO2 SCR Reaction. For a realistic simulation of the NO2 SCR reaction, NH3 adsorption/desorption and NH3 oxidation should be considered together. Therefore, in this study, kinetic parameter tuning for the NO2 SCR reaction was conducted using the already-estimated kinetics of NH3 adsorption/desorption and NH3 oxidation.

Figure 17. Comparisons of calculated and measured exit concentrations of (a) NO (b) NO2, and (c) NH3 for the steady-state NOx SCR reaction. Space velocity = 30000 h-1.

On the basis of the currently developed numerical tools, kinetic parameter calibration for the NO2 SCR reaction was carried out using the steady-state experimental results given in Figure 4. In this calibration, the data measured at 30000, 40000, and 50000 h-1 were simultaneously employed. However, measurement data showing 100% conversion were excluded.
2860
dx.doi.org/10.1021/ie101558d |Ind. Eng. Chem. Res. 2011, 50, 28502864

Industrial & Engineering Chemistry Research

ARTICLE

Figure 18. Comparisons of calculated and measured exit concentrations of (a) NO, (b) NO2, and (c) NH3 for the steady-state NOx SCR reaction. Space velocity = 40000 h-1.

Figure 19. Comparisons of calculated and measured exit concentrations of (a) NO, (b) NO2, and (c) NH3 for the steady-state NOx SCR reaction. Space velocity = 50000 h-1.

Figure 12 shows an Arrhenius plot for the newly tted preexponential factors of the NO2 SCR reaction rate. To obtain each tted data point presented in Figure 12, an iterative calculation was performed while updating the pre-exponential factor. The stopping criterion was that the deviation between the calculated and measured species concentrations at the SCR exit became as small as 0.1 ppm. Here, both NO2 and NH3 were employed for stopping criteria, so that two sets of tted data were produced at one catalyst temperature. Analyzing the Arrhenius plot with reference to Wang et al.32 gave a new pre-exponential factor and activation energy of the NO2 SCR reaction rate, as summarized in Table 4. The initial kinetic parameters employed for this calibration were taken from Olsson et al.19 and are also listed in Table 4. Figures 13-15 show comparisons of the calculated and measured results for NO2 SCR reaction at space velocities of 30000, 40000, and 50000 h-1, respectively. At 30000 h-1, quite good agreement is observed for both species. However, the deviation between simulation and experiment becomes larger as space velocity increases, especially at 50000 h-1. NOx SCR Reaction. For a realistic simulation of the NOx SCR reaction, NH3 adsorption/desorption, NH3 oxidation, NO SCR, and NO2 SCR should be considered together. Therefore, in this study, kinetic parameter tuning for the NOx SCR reaction was conducted using the already-estimated kinetics of NH3 adsorption/desorption, NH3 oxidation, NO SCR, and NO2 SCR. On the basis of the currently developed numerical tools, kinetic parameter calibration for the NOx SCR reaction was performed using the steady-state experimental results given in Figure 5. In this calibration, all of the data measured at 30000,

40000, and 50000 h-1 were simultaneously utilized. However, measurement data showing 100% conversion were excluded because they are not useful in estimating the kinetic parameters as described previously. In the end, only the following lowtemperature data were employed in the calibration: two data points below 269 C at 30000 h-1, three data points below 304 C at 40000 h-1, and three data points below 298 C at 50000 h-1. Figure 16 displays an Arrhenius plot for the newly tted preexponential factors of the NOx SCR reaction rate. To obtain each tted data point presented in Figure 16, an iterative calculation was performed while updating the pre-exponential factor. The stopping criterion was that the deviation between the calculated and measured exit concentrations became as small as 0.1 ppm. Here, both NO and NH3 were used for stopping criteria, so that two sets of tted data were produced at one catalyst temperature. Note that NO2 was not selected for a stopping criterion because all of the measurement data for NO2 showed 100% conversion. Analyzing the Arrhenius plot with reference to Wang et al.32 yielded a new pre-exponential factor and activation energy, as summarized in Table 4. The initial kinetic parameters used in this calibration were taken from Olsson et al.19 and are also listed in Table 4. Figures 17-19 compare the calculated and measured results for the NOx SCR reaction at space velocities of 30000, 40000, and 50000 h-1, respectively. For NH3, good agreement is observed between the simulation and experiment at all space velocities. However, for NO and NO2, the simulation results show some deviations from the experimental results, which becomes larger as space velocity increases. Especially for NO2,
2861
dx.doi.org/10.1021/ie101558d |Ind. Eng. Chem. Res. 2011, 50, 28502864

Industrial & Engineering Chemistry Research

ARTICLE

Figure 20. Variations in the SCR inlet concentrations of NO, NO2, and NH3 with time for both reactor temperatures of 230 and 300 C.

Figure 22. Comparisons of calculated and measured exit concentrations of (a) NO, (b) NO2, and (c) NH3 for the transient NOx SCR reaction at a constant reactor temperature of 300 C. Space velocity = 40000 h-1.

Table 5. Activation Energy of Each Reaction Rate Yielding the Best Fit of the Simulation to Transient Experiments at Constant Reactor Temperature
activation energy (kJ/mol) reaction 230 C 125.3 81.4 48.3 68.4 300 C 125.3 80.9 50.8 68.8

Figure 21. Comparisons of calculated and measured exit concentrations of (a) NO, (b) NO2, and (c) NH3 for the transient NOx SCR reaction at a constant reactor temperature of 230 C. Space velocity = 40000 h-1.

NH3 oxidation NO SCR NO2 SCR NOx SCR

the discrepancy becomes considerable in both the low- and hightemperature regions.

5. TRANSIENT VALIDATION OF THE MODEL For the NOx SCR reaction, a validation of the transient simulation results produced using the currently obtained kinetic parameters and computational code was performed with the transient experimental results. Figure 20 displays the transient SCR inlet conditions for both reactor temperatures of 230 and 300 C. The total measurement duration was 2430 s for both cases. In Figures 21 and 22, the transient simulation results are compared with the experimental

results, and it can be seen that there is quite good agreement. However, for the NO exit concentration, some deviations can be observed in the early part of the transient process, especially at 300 C. Also, for the NH3 exit concentration, the simulation results exhibit a smooth nature in their transient variations, whereas the experimental results show a discrete characteristic. In the transient simulations, the activation energy of each reaction rate was changed slightly from the values presented in Table 4 for the best t to the experimental results. This tuning was carried out on the basis of the steady-state zone of the transient experiments. The currently calibrated activation energies are given in Table 5. Note that a time increment of 1 s was employed in the current transient calculations.
2862
dx.doi.org/10.1021/ie101558d |Ind. Eng. Chem. Res. 2011, 50, 28502864

Industrial & Engineering Chemistry Research

ARTICLE

6. CONCLUSIONS In this study, an in-house computational code for simulating the performance of ammonia/urea SCR was developed. On the basis of this numerical tool, kinetic parameter calibrations for various catalytic reactions were successfully conducted using the steady-state experimental results obtained for a commercial ammonia/urea SCR washcoated Fe-ion-exchanged zeolite-based catalyst. Also, the transient simulation results were validated with experimental results. A summary of the main results of this study is as follows:
1. The NH3 TPD experimental results are nicely described by the simulation results obtained using the newly calibrated kinetic parameters. Both adsorption and temperature-programmed desorption of NH3 over SCR catalyst surface are well-predicted. 2. At a space velocity of 10000 h-1, NH3 oxidation with O2 began at about 350 C, and the conversion rate reached nearly 60% at 500 C. For steady-state NH3 oxidation, the simulation results obtained using the newly calibrated kinetic parameters showed quite good agreement with the experimental results. 3. In the NO SCR experiments, complete NO conversion was maintained up to catalyst temperatures as high as 500 C at a space velocity of 10000 h-1. For the steady-state NO SCR reaction, the simulation results obtained using the newly calibrated kinetic parameters displayed excellent agreement with the experimental results. 4. At space velocities of 30000 to 50000 h-1, NO2 removal through the NO2 SCR reaction was found to be enhanced with increasing catalyst temperature up to around 350 C. Above this temperature range, the NO2 conversion rate decreased again because of the lack of NH3 by its oxidation with O2. For the steady-state NO2 SCR reaction, the simulation results produced using the newly estimated kinetic parameters closely followed the experimental results; however, the simulation error became larger as space velocity increased. 5. From the steady-state NOx SCR experiments, the consumption ratio of NOx, (NO NO2)/NH3, appeared to be almost 1:1. Also, it was observed that the conversion rate of NO2 was higher than that of NO. For the NH3 concentration, the simulation results obtained using the newly estimated kinetic parameters nicely predicted the experimental results. However, for the NO and NO2 concentrations, some deviations were found between the simulations and the experiments. 6. Transient NOx SCR reaction processes at constant reactor temperatures of 230 and 300 C were well-predicted by the currently obtained kinetic parameters and numerical code.

ACKNOWLEDGMENT This work was supported by a Korea Science and Engineering Foundation (KOSEF) grant funded by the Korean government (MEST) (2010-0000353). NOMENCLATURE ac = catalytic active-site density (mol/m3) Ci = molar concentration of species i (mol/m3) Ea = activation energy of NH3 adsorption (J/mol) Ed = activation energy of NH3 desorption ( J/mol) Eo = activation energy of NH3 desorption at zero coverage d ( J/mol) ka = rate constant of NH3 adsorption [m3/(mol s)] ko = pre-exponential factor of NH3 adsorption rate constant a [m3/(mol s)] kd = rate constant of NH3 desorption (s-1) ko = pre-exponential factor of NH3 desorption rate constant d (s-1) ki = rate constant of reaction i [m3/(mol s)] ko = pre-exponential factor of rate of reaction i [m3/(mol s)] i Ra = NH3 adsorption rate (s-1) Rd = NH3 desorption rate (s-1) RNO = NO SCR reaction rate (s-1) Rox = NH3 oxidation rate (s-1) Ru = universal gas constant [ J/(mol K)] t = time (s) T = temperature (K) uD = supercial velocity (m/s) x = axial position (m) Y = convergence criterion variable
Greek Symbols

R = parameter for the dependence of the ammonia surface coverage = iterative error = void fraction (macroscopic bulk porosity) NH3 = ammonia surface coverage F = density (kg/m3)
Subscripts

a = NH3 adsorption b = bulk d = NH3 desorption g = gas phase i = dummy index m = monolith p = pore, packed-bed s = solid phase
Superscripts

AUTHOR INFORMATION
Corresponding Author

n = iterative index

*Tel.: 82-31-270-1378 (T.J.W.). Fax: 82-31-270-1399 (T.J.W.). E-mail: cakecandle@kaist.ac.kr (T.J.W.), isnam@ postech.ac.kr (I.-S.N.).
Present Addresses

REFERENCES
(1) Selective Catalytic Reduction. In DieselNet Technology Guide. Ecopoint Inc.: Mississauga, Ontario, Canada, 2002. Available by subscription at http://www.dieselnet.com/tech/cat_scr.html (Accessed July 2010). (2) Koebel, M.; Elsener, M.; Kleemann, M. Urea-SCR: A Promising Technique to Reduce NOx Emissions from Automotive Diesel Engines. Catal. Today 2000, 59, 335345. (3) Yim, S. D.; Kim, S. J.; Baik, J. H.; Nam, I. S.; Mok, Y. S.; Lee, J. H.; Cho, B. K.; Oh, S. H. Decomposition of Urea into NH3 for the SCR Process. Ind. Eng. Chem. Res. 2004, 43, 48564863.
2863
dx.doi.org/10.1021/ie101558d |Ind. Eng. Chem. Res. 2011, 50, 28502864

Advanced Combustion & Engine Research Team, Institute of Technology, Doosan Infracore, 39-3 Sungbok-dong, Suji-gu, Yongin-si, Gyeonggi-do 448-795, Korea.

Energy Laboratory, Samsung Advanced Institute of Technology (SAIT), Samsung Electronics Co., Ltd., San 14, Nongseodong, Giheung-gu, Yongin-si, Gyeonggi-do 446-712, Korea.

Industrial & Engineering Chemistry Research

(4) Seker, E.; Yasyerli, N.; Gulari, E.; Lambert, C.; Hammerle, R. H. NOx Reduction by Urea under Lean Conditions over Single Step Sol-Gel Pt/Alumina Catalyst. Appl. Catal. B: Environ. 2002, 37, 2735. (5) Fang, H. L.; DaCosta, H. F. M. Urea Thermolysis and NOx Reduction with and without SCR Catalysts. Appl. Catal. B: Environ. 2003, 46, 1734. (6) Narayanaswamy, K.; He, Y. Modeling of Copper-Zeolite and Iron-Zeolite Selective Catalytic Reduction (SCR) Catalysts at Steady State and Transient Conditions; SAE Paper 2008-01-0615; SAE International: Warrendale, PA, 2008. (7) Girard, J. W.; Montreuil, C.; Kim, J.; Cavataio, G.; Lambert, C. Technical Advantages of Vanadium SCR Systems for Diesel NOx Control in Emerging Markets; SAE Paper 2008-01-1029; SAE International: Warrendale, PA, 2008. (8) Cavataio, G.; Girard, J.; Patterson, J. E.; Montreuil, C.; Cheng, Y.; Lambert, C. K. Laboratory Testing of Urea-SCR Formulations to Meet Tier 2 Bin 5 Emissions; SAE Paper 2007-01-1575; SAE International: Warrendale, PA, 2007. (9) Johnson, T. V. Diesel Emission Control in Review; SAE Paper 2009-01-0121; SAE International: Warrendale, PA, 2009. (10) Fedeyko, J. M.; Chen, H. Y.; Ballinger, T. H.; Weigert, E. C.; Chang, H. L.; Cox, J. P.; Andersen, P. J. Development of Thermally Durable Cu/SCR Catalysts; SAE Paper 2009-01-0899; SAE International: Warrendale, PA, 2009. (11) Giesho, J.; Schfer-Sindlinger, A.; Spurk, P. C.; Tillaart, a J. A. A.; Garr, G. Improved SCR Systems for Heavy Duty Applications; SAE Paper 2000-01-0189; SAE International: Warrendale, PA, 2000. (12) Girard, J.; Cavataio, G.; Snow, R.; Lambert, C. Combined Fe-Cu SCR Systems with Optimized Ammonia to NOx Ratio for Diesel NOx Control; SAE Paper 2008-01-1185; SAE International: Warrendale, PA, 2008. (13) Cavataio, G.; Jen, H. W.; Warner, J. R.; Girard, J. W.; Kim, J. Y.; Lambert, C. K. Enhanced Durability of a Cu/Zeolite Based SCR Catalyst; SAE Paper 2008-01-1025; SAE International: Warrendale, PA, 2008. (14) Felix, S. P.; Savill-Jowitt, C.; Brown, D. R. Base Adsorption Calorimetry for Characterising Surface Acidity: A Comparison between Pulse Flow and Conventional Static Techniques. Thermochim. Acta 2005, 433, 5965. (15) Komatsu, T.; Nunokawa, M.; Moon, I. S.; Takahara, T.; Namba, S.; Yashima, T. Kinetic Studies of Reduction of Nitric Oxide with Ammonia on Cu2-Exchanged Zeolites. J. Catal. 1994, 148, 427 437. (16) Olsson, L.; Sjvall, H.; Blint, R. J. Detailed Kinetic Modeling of o NOx Adsorption and NO Oxidation over Cu-ZSM-5. Appl. Catal. B: Environ. 2009, 87, 200210. (17) Wurzenberger, J. C.; Wanker, R. Multi-Scale SCR Modeling, 1D Kinetic Analysis and 3D System Simulation; SAE Paper 2005-01-0948; SAE International: Warrendale, PA, 2005. (18) Chatterjee, D.; Burkhardt, T.; Weibel, M.; Nova, I.; Grossale, A.; Tronconi, E. Numerical Simulation of Zeolite- and V-Based SCR Catalytic Converters; SAE Paper 2007-01-1136; SAE International: Warrendale, PA, 2007. (19) Olsson, L.; Sjvall, H.; Blint, R. J. A Kinetic Model for Ammonia o Selective Catalytic Reduction over Cu-ZSM-5. Appl. Catal. B: Environ. 2008, 81, 203217. (20) Schuler, A.; Votsmeier, M.; Malmberg, S.; Giesho, J.; Drochner, A.; Vogel, H.; Kiwic, P. Dynamic Model for the Selective Catalytic Reduction of NO with NH3 on Fe-Zeolite Catalysts; SAE Paper 2008-01-1323; SAE International: Warrendale, PA, 2008. (21) Tronconi, E.; Lietti, L.; Forzatti, P.; Malloggi, S. Experimental and Theoretical Investigation of the Dynamics of the SCR - DeNOx Reaction. Chem. Eng. Sci. 1996, 51 (11), 29652970. (22) Lietti, L.; Nova, I.; Camurri, S.; Tronconi, E.; Forzatti, P. Dynamics of the SCR-DeNOx Reaction by the Transient-Response Method. AIChE J. 1997, 43 (10), 25592570. (23) Schaub, G.; Unruh, D.; Wang, J.; Turek, T. Kinetic Analysis of Selective Catalytic NOx Reduction (SCR) in a Catalytic Filter. Chem. Eng. Process 2003, 42, 365371.

ARTICLE

(24) Koebel, M.; Elsener, M.; Madia, G. Recent Advances in the Development of Urea-SCR for Automotive Applications; SAE Paper 200101-3625; SAE International: Warrendale, PA, 2001. (25) Baik, J. H.; Roh, J. H.; Yim, S. D.; Nam, I. -S.; Lee, J. -H.; Cho, B. K.; Oh, S. H. Synergistic Roles of NO and NO2 in Selective Catalytic Reduction of NOx by NH3. Stud. Surf. Sci. Catal. 2006, 159, 441444. (26) Devadas, M.; Krcher, O.; Elsener, M.; Wokaun, A.; Sger, N.; o o Pfeifer, M.; Demel, Y.; Mussmann, L. Inuence of NO2 on the Selective Catalytic Reduction of NO with Ammonia over Fe-ZSM5. Appl. Catal. B: Environ. 2006, 67, 187196. (27) Kato, A.; Matsuda, S.; Kamo, T.; Nakajima, F.; Kuroda, H.; Narita, T. Reaction between NOx and NH3 on Iron Oxide-Titanium Oxide Catalyst. J. Phys. Chem. 1981, 85, 40994102. (28) Tuenter, G.; Leeuwen, W. F.; Snepvangers, L. J. M. Kinetics and Mechanism of the NOx Reduction with NH3 on V2O5-WO3-TiO2 Catalyst. Ind. Eng. Chem. Prod. Res. Dev. 1986, 25 (4), 633636. (29) Stevenson, S. A.; Vartuli, J. C.; Sharma, S. B. The Eects of Steaming and Sodium Exchange on the Selective Catalytic Reduction of NO and NO2 by NH3 over HZSM-5. J. Catal. 2002, 208, 106113. (30) Koebel, M.; Elsener, M.; Madia, G. Reaction Pathway in the Selective Catalytic Reduction Process with NO and NO2 at Low Temperatures. Ind. Eng. Chem. Res. 2001, 40, 5259. (31) Tannehill, J. C.; Anderson, D. A.; Pletcher, R. H. Computational Fluid Mechanics and Heat Transfer, 2nd ed.; Taylor and Francis: Philadelphia, PA, 1997. (32) Wang, T. J.; Baek, S. W.; Lee, J. H. Kinetic Parameter Estimation of a Diesel Oxidation Catalyst under Actual Vehicle Operating Conditions. Ind. Eng. Chem. Res. 2008, 47, 25282537.

2864

dx.doi.org/10.1021/ie101558d |Ind. Eng. Chem. Res. 2011, 50, 28502864