A Predictive Mechanism for Mercury Oxidation on Selective Catalytic ...

11 downloads 50 Views 245KB Size Report
oxidation by NH3 for NH3/NO ratios from zero to 0.9. Perhaps most important for ... FGDs.3,4 But specialized pretreatment will usually not be necessary when the ...
TECHNICAL PAPER

ISSN 1047-3289 J. Air & Waste Manage. Assoc. 55:1866 –1875 Copyright 2005 Air & Waste Management Association

A Predictive Mechanism for Mercury Oxidation on Selective Catalytic Reduction Catalysts under Coal-Derived Flue Gas Stephen Niksa Niksa Energy Associates, Belmont, CA Naoki Fujiwara Coal Research Laboratory, Idemitsu Kosan Co., Ltd., Chiba, Japan

ABSTRACT This paper introduces a predictive mechanism for elemental mercury (Hg0) oxidation on selective catalytic reduction (SCR) catalysts in coal-fired utility gas cleaning systems, given the ammonia (NH3)/nitric oxide (NO) ratio and concentrations of Hg0 and HCl at the monolith inlet, the monolith pitch and channel shape, and the SCR temperature and space velocity. A simple premise connects the established mechanism for catalytic NO reduction to the Hg0 oxidation behavior on SCRs: that hydrochloric acid (HCl) competes for surface sites with NH3 and that Hg0 contacts these chlorinated sites either from the gas phase or as a weakly adsorbed species. This mechanism explicitly accounts for the inhibition of Hg0 oxidation by NH3, so that the monolith sustains two chemically distinct regions. In the inlet region, strong NH3 adsorption minimizes the coverage of chlorinated surface sites, so NO reduction inhibits Hg0 oxidation. But once NH3 has been consumed, the Hg0 oxidation rate rapidly accelerates, even while the HCl concentration in the gas phase is

uniform. Factors that shorten the length of the NO reduction region, such as smaller channel pitches and converting from square to circular channels, and factors that enhance surface chlorination, such as higher inlet HCl concentrations and lower NH3/NO ratios, promote Hg0 oxidation. This mechanism accurately interprets the reported tendencies for greater extents of Hg0 oxidation on honeycomb monoliths with smaller channel pitches and hotter temperatures and the tendency for lower extents of Hg0 oxidation for hotter temperatures on plate monoliths. The mechanism also depicts the inhibition of Hg0 oxidation by NH3 for NH3/NO ratios from zero to 0.9. Perhaps most important for practical applications, the mechanism reproduces the reported extents of Hg0 oxidation on a single catalyst for four coals that generated HCl concentrations from 8 to 241 ppm, which covers the entire range encountered in the U.S. utility industry. Similar performance is also demonstrated for full-scale SCRs with diverse coal types and operating conditions.

IMPLICATIONS Companies contemplating Hg control with wet flue gas desulfurization (FGD) need to know in advance how much of the Hg at the FGD inlet is oxidized, because only oxidized Hg is soluble in scrubber solutions. SCR catalysts for nitrogen oxides (NOX) control effectively promote upstream Hg oxidation, provided that sufficient HCl is present and that the SCR provides sufficient residence time and surface area for the Hg oxidation to counteract its inhibition by the catalytic NO reduction chemistry. The combined reaction and transport analysis introduced here accurately predicts the extents of Hg oxidation for the pertinent ranges of coal quality and SCR conditions, including variable HCl levels, NH3/NO ratios, space velocities, temperatures, monolith pitches, and channel shapes. Companies can use the analysis to forecast Hg oxidation in their existing or contemplated SCRs for any coal type over the practical range of SCR operating conditions.

INTRODUCTION The coal-burning utility industry is beginning to mount a massive response to comply with impending regulations on mercury (Hg) emissions from power plants, and a variety of control strategies are now under intensive development. The most economic approaches use existing particulate control devices and flue gas desulfurization (FGD) scrubbers, often in conjunction with additives, sorbents, catalytic pretreatment reactors, and other supplemental measures to enhance Hg recoveries. Any control strategy that retains Hg in FGD scrubber solutions and solids can only be as effective as the proportion of oxidized Hg at the FGD inlet, because oxidized vapor species (Hg[II]), particularly mercuric chloride (HgCl2), are watersoluble and, therefore, are easily dissolved in scrubbing solutions, whereas elemental (Hg0) vapor is insoluble and, therefore, is able to pass through scrubbers into the smokestack. Consequently, factors that promote Hg0

1866 Journal of the Air & Waste Management Association

Volume 55 December 2005

Niksa and Fujiwara oxidation upstream of the FGD are essential components of this approach. Hg0 oxidizes in the vapor phase in the presence of chlorine (Cl) atoms and Cl2 but at rates that are almost always much slower than its heterogeneous oxidation on unburned carbon particles (UBCs) in the fly ash.1,2 So heterogeneous Hg0 oxidation on UBC is usually the fastest inherent mechanism to oxidize Hg0 in coal-derived flue gas. Unfortunately, even for the relatively high levels of UBC and Cl species associated with Eastern high-volatile (hv) bituminous coals, the inherent Hg0 oxidation mechanism rarely delivers ⬎70% Hg(II) to the FGD inlet.2 The proportions of Hg(II) are significantly lower with subbituminous coals and lignites and with many hv bituminous coals. Catalytic Hg0 oxidizers are being developed to boost the levels of Hg(II) species immediately upstream of FGDs.3,4 But specialized pretreatment will usually not be necessary when the gas cleaning system contains selective catalytic reduction (SCR) catalysts for nitrogen oxides (NOX) control. SCR catalysts effectively oxidize Hg0, provided that sufficient HCl is present and that the SCR provides sufficient residence time and surface area for the Hg oxidation to counteract its inhibition by the catalytic nitric oxide (NO) reduction chemistry. The broad range of Hg0 oxidation performance monitored across full-scale SCRs5–7 is consistent with the wide variations in coal Cl and SCR operation across American coal-burning utilities. Generally speaking, coals with more chlorine exhibit higher Hg0 oxidation performance on SCRs. But this tendency is frequently counteracted by unfavorable SCR operation, because of inadequate residence times, surface areas, and excessive inhibition from the NO reduction chemistry. This paper analyzes the transport and chemistry along SCR catalysts to relate coal quality and SCR operating conditions to Hg0 oxidation performance, given the Hg speciation and HCl concentration at the SCR inlet, and the SCR temperature, space velocity, NH3/NO ratio, monolith pitch, and channel shape. Evaluations with the available laboratory-scale and pilot-scale database demonstrate the capability to quantitatively depict the impact of coal Cl and all of the important SCR operating characteristics. Evaluations with full-scale SCR test data demonstrate satisfactory performance in commercial applications. Based on this performance, the analysis is ultimately used to rank order the major factors affecting Hg0 oxidation on SCR catalysts with parametric sensitivity studies. Reaction Mechanism and Transport Analysis Throughout the 1990s, several research groups characterized the mechanisms for NO reduction and simultaneous Volume 55 December 2005

sulfur dioxide (SO2) oxidation on commercial vanadium pentoxide(V2O5)/tungsten trioxide (WO3) on titanium dioxide (TiO2) SCR catalysts. The mechanism is now understood in sufficient detail8 to interpret the impacts of all of the major operating conditions in terms of the following experimental observations: (1) the primary acidic sites involve V2O5, although additional weaker sites involve WO3, TiO2, and surface sulfates; (2) no NO is observed on the catalyst surface, so NO reacts as either a very weakly adsorbed surface species or directly from the gas phase; (3) ammonia (NH3) strongly adsorbs on both Lewis and Brønsted acid sites; (4) NO is reduced by NH3 adsorbed onto V2O5 sites, and sites of weaker adsorbed NH3 act as a reservoir to replenish the strongly adsorbed NH3; and (5) reduced surface sites are oxidized by O2 in a redox cycle. According to this mechanism, the NO reduction rate is affected by the NH3/NO ratio, as well as the concentrations of SO2, O2, and H2O. Whereas NH3/NO is obviously important, the other gas species are inconsequential at their concentrations in coal-derived flue gas.9,10 Once sulfates have been deposited onto the catalyst surface by exposure to SO2/SO3, the NO reduction rate is independent of the SO2 concentration. Similarly, O2 and H2O do not affect the NO reduction rate at concentrations of a few percent or more, so these species are inconsequential in all coal-derived flue gas. Utility SCRs always operate with NH3/NO ratios below unity to meet the regulatory limit that NH3 slip remains ⬍2 ppm. So NH3 is the deficient reactant of which the concentration necessarily decays from the inlet value to essentially zero. The following Eley-Rideal rate law is most commonly used for this situation:11,12

KNH3CNH3 dCNO C ⫽ kNO dt 1 ⫹ KNH3CNH3 NO

r NO ⫽ ⫺

(1)

where kNO is a global surface rate constant, and KNH3 is the NH3 adsorption equilibrium constant. Initially, when CNH3 is comparable with CNO, the rate is independent of CNH3. But as NH3 is depleted, the rate becomes proportional to CNH3. The NO reduction rate is first-order in the NO concentration over the entire range of NH3 concentrations. NO reduction on commercial SCR catalysts is definitely not a kinetically limited process, so the rate law in eq 1 must be combined with interphase and intraphase transport mechanisms to quantitatively analyze the performance of commercial SCR systems. Film transport coefficients have been characterized in detail for monoliths with square, circular, and triangular channels9,11,12 and for the more complex plate-type monoliths.13,14 From a rigorous perspective, the film transport coefficients are Journal of the Air & Waste Management Association 1867

Niksa and Fujiwara functions of the distance from the inlet, as well as the surface reaction rate. However, Tronconi et al.11,12 demonstrated that mass transfer coefficients may be evaluated from the familiar Graetz-Nusselt analysis of heat transfer across developing laminar boundary layers in confined channels, without consideration of the surface chemistry. In particular, they demonstrated that a one-dimensional analysis with transport coefficients based on the local Nusselt number for a constant wall temperature closely matched the results from the much more complex twodimensional treatment.11,12 The recommended expression is as follows: Sh ⫽

k m DH ⫽ Nu⬁,T ⫹ 8.827共1000Z*兲 ⫺ 0.545 exp共⫺48.2Z*兲 DNO (2)

where Z* is the Graetz axial coordinate for mass transfer, z/ReSc; Re is the Reynolds number, ␳uDH/␮; ␳, ␮, u, and D4 are the bulk gas density, viscosity, flow velocity, and hydraulic diameter, respectively; Sc is the Schmidt number, ␮/␳DNO; and DNO is the diffusivity of NO. The transport coefficient relaxes from a very large value at the SCR inlet to the asymptotic value of the Nusselt number for an isothermal wall, Nu⬁,T. The asymptote varies for channels of different shapes. The same approach was also demonstrated for plate-type monoliths by using a weighted average of the coefficients for rectangular and triangular channels to represent the irregular flow cross-sections.14 We will expand the established mechanism for catalytic NO reduction to include Hg0 oxidation by resolving film transport from lumped surface reactivities for NO and Hg0. Note, however, that it has already been established that pore diffusion within commercial SCR catalysts significantly mediates the overall NO reduction rates. Typical effectiveness factors are ⬍0.1.9,11,12 Pore diffusion mechanisms must be added to support catalyst optimization work but are not necessary to relate coal quality impacts and SCR operating conditions to the Hg0 oxidation performance. Hence, previous research on NO reduction over commercial SCR catalysts establishes the following: (1) the form of the rate expression for NO reduction; and (2) the necessary coefficients for film transport. The means to resolve pore transport effects has also been reported9,11 but will be relegated to a future expansion for catalyst design applications. The most important finding to connect NO reduction and Hg0 oxidation is the inverse relation between the conversion of NO and Hg0 on a conventional V2O5 on TiO2 honeycomb, which is reproduced in Figure 2, below. As the molar NH3/NO ratio was increased from zero to unity to span the complete range of NO reduction efficiencies, the extent of Hg oxidation fell progressively 1868 Journal of the Air & Waste Management Association

from 82 to 39%.15 Machalek et al.15 also showed that NH3 injection displaces the relation between Hg0 oxidation and space velocity downward so that lower extents of Hg0 oxidation are achieved at every space velocity during NH3 injection (compare with Figure 3, below). Also, in all of the full-scale tests in which the SCR was operated with and without NH3 injection, NH3 injection inhibited Hg0 oxidation. We regard these findings as direct evidence for competitive adsorption between NH3 and an active Hg0 oxidation agent, which is probably HCl.16 Our main underlying premise is that SCR catalysts contain sites that can adsorb both NH3 and HCl. Hence, a competitive adsorption process relates Hg0 oxidation to NO reduction. Chlorinated sites sustain Hg0 oxidation, whereas ammoniated sites sustain NO reduction. This competition will be described with adsorption equilibrium constants for both species. Note also that the rate of Hg0 oxidation is far too slow to perturb the HCl concentration across the SCR because of the disparate inlet concentrations of HCl and Hg0. This implies that the HCl concentration in the gas phase is uniform across the SCR. However, the NH3 concentration falls continuously with distance along the monolith and, according to the postulated NH3 adsorption equilibrium, the surface coverage of adsorbed NH3 diminishes accordingly. Although the HCl vapor concentration is uniform, the surface coverage of chlorinated sites grows, because there are fewer adsorbed NH3 species to compete for sites further into the channels. Hence, the impact of NH3 inhibition segregates the catalytic monolith into an upstream section where Hg0 oxidation is strongly inhibited by the high NH3 concentration and a downstream section where NH3 inhibition eventually vanishes. Film transport resistances for Hg0 must be at least as important as they are for NO because the reported length scales for Hg0 oxidation are comparable with those for NO reduction,17 and the diffusivity of Hg0 is lower than the diffusivity of NO because of its greater molecular weight. Therefore, in the steady state, the fluxes of NO and Hg0 onto the catalytic channel walls must balance their overall consumption because of catalytic conversion, as follows:

S S 兲 ⫽ kNOCNO ⌰NH3 k m,NO 共CNO ⫺ CNO

⫽ kNO

S KNH3CNH 3 0 S 1 ⫹ KNH3CNH ⫹ KHClCHCl 3

S CNO

(3)

k m,Hg 共CHg ⫺ CSHg兲 ⫽ kHgCSHg⌰Cl ⫽ kHg

0 KHClCHCl CSHg 0 S 1 ⫹ KNH3CNH ⫹ KHClCHCl 3

(4)

Volume 55 December 2005

Niksa and Fujiwara where km,i is the film transport coefficient of either NO or Hg0; Ci are molar species concentrations where superscript S denotes conditions immediately above the surface and superscript 0 denotes inlet conditions in the bulk flue gas; ki is the lumped surface reactivity; ⌰i are fractional surface coverages of NH3 and chloride; and Ki are adsorption equilibrium constants for NH3 and HCl. The EleyRideal rate laws were derived for competitive adsorption equilibrium of NH3 and HCl. In the steady state, the molar concentrations change while the flow moves through the catalytic channel because of transport of reactants that sustain the surface reactions, according to the following equation:

u

dC i ⫺4km,i 共Ci ⫺ CiS兲 ⫽ dz DH

where

DH ⫽

4AX P

(5)

where z is the axial position. In the definition of the hydraulic diameter, DH, AX is the flow cross-section, and P is the wetted perimeter. Versions of eq 5 for NO and Hg0 are solved subject to specified NO and Hg0 concentrations at the monolith inlet. It is not necessary to include a species conservation balance for NH3, because the equivalent stoichiometries for both reactants during NO reduction match both their concentration gradients along the monolith and also their transport fluxes onto the wall at each axial position. These two conditions determine CNH3 and CSNH3 along the channel from the corresponding NO concentrations. On a P4 microprocessor operating at 1.5 GHz, each numerical solution for CNO and CHG0 with an Adams-Gear routine for stiff coupled ODEs takes ⬍1 s. Four rate parameters need to be specified: KNH3, kNO, KHCl, and kHg. All are of the Arrhenius form. The intrinsic activation energy for NO reduction is 20 –25 kcal/mol, but the apparent activation energies reported for commercial SCR catalysts are 5–10 kcal/mol18 because of strong influences of pore size and site concentrations. The assigned values for the NH3 adsorption equilibrium constant at 350 °C vary by a factor of 300.9,14,19 –21 It has been asserted11,12 that this “constant” should vary widely for different catalysts, because the adsorption equilibrium is governed by the number and strength of acidic sites. In turn, these sites are affected by temperature, the coverage of surface sulfates, the concentrations of NH3 and SO2, and the SO3 conversion efficiency. The reported scaling for temperature is also highly variable, ranging from no temperature dependence11,12 to a heat of adsorption of 38 kcal/mol.19 All of the parameters associated with NO reduction must, therefore, be regarded as adjustable, except that the apparent activation energy for NO reduction should be between 5 and 10 kcal/mol. Moreover, no measurements have yet been reported to estimate any of the parameters in kHg or KHCl. Volume 55 December 2005

Input Requirements This mechanism requires the following input data: (1) C0NO, C0Hg0, and C0HCl (none of the other major gas components need to be specified provided that the O2 and H2O levels exceed a few percent); (2) the total Hg concentration and the fraction of Hg(II) at the SCR inlet; (3) the molar NH3/NO ratio; (4) The SCR temperature or an axial temperature profile if the temperature gradient exceeds 30 °C; (5) the gas hourly space velocity (GHSV), evaluated at 0 °C; (6) the monolith configuration (plate versus honeycomb) and the pitch (i.e., unit cell dimension in the monolith) and channel shape; and (7) the catalyst manufacturer, to provide a basis for clarifying geometric specifications and for estimating rate parameters. Monolith pitch is the characteristic dimension of the unit cell. It is easiest to evaluate as the sum of the wall thickness and the opening size. Based on reported wall thicknesses from manufacturer websites, wall thicknesses are typically 10% of the pitch. Such values typically yield flow voidages approaching 80% and geometric surface areas from 300 to 900 m2/m3. The catalyst manufacturer should be specified so that the assigned reactivity parameters for similar formulations may be compared for consistency. Ultimately, one hopes that accurate default values may be identified for each manufacturer, because the active metal formulations are proprietary. When data are to be used to evaluate the predicted extents of Hg0 oxidation across SCR catalysts, measured extents of Hg0 oxidation at the SCR outlet are required. It is also useful to have NH3 slip measurements, because the analysis also predicts the NO reduction efficiency. SCR Database Among the reported datasets that include all of the required input, datasets from two laboratory-scale studies16,22 and one slipstream facility15 more than cover the practical ranges of HCl concentration, NH3/NO ratio, temperature, space velocity, and monolith specifications, including honeycomb versus plate configurations. A second phase of the evaluations emphasizes the impact of coal quality and catalyst type on the performance of pilotscale and full-scale SCRs. Bock et al.22 reported an appreciable reduction in the extent of Hg0 oxidation over a plate catalyst when the HCl concentrations were reduced from 60 to 10 ppm, consistent with the poor oxidation performance of commercial SCR units with flue gas derived from subbituminous coals, compared with bituminous-derived flue gas. They also resolved Hg0 oxidation across two square-channel honeycombs and one plate monolith that had the same geometric surface-to-volume ratios, thereby isolating the impact of cell pitch and channel shape on the Hg0 oxidation rate. Whereas both honeycombs may have had Journal of the Air & Waste Management Association 1869

Niksa and Fujiwara the same catalyst formulation, the plate monolith had a different formulation, because the honeycombs were extruded but the plate monolith was wash-coated. Accordingly, the same rate parameters were assigned for both honeycomb monoliths, but different parameters were applied to the plate monolith. Bock et al.22 also characterized the temperature dependences from 280 to 420 °C on all three of the monoliths. Machalek et al.15 pulled subbituminous-derived flue gas into a pilot-scale SCR from a full-scale gas cleaning system, finding that extents of Hg0 oxidation decreased from 40 to 25 to 5% as space velocity was increased from 3000 to 5500 to 7800 hr⫺1. These results stayed the same over 3200 hr of testing. However, with NH3 injection, the catalyst completely deactivated for Hg0 oxidation after 4200 hr of exposure, and the initial 85% of Hg0 oxidation completely vanished. Whereas most operating conditions were reported, coal properties had to be estimated from the nominal subbituminous coal properties for the same power plant that were reported in another Hg testing program. The only datasets on Hg transformations across fullscale SCRs with most of the required characterization data were recorded by the testing team at the University of North Dakota Energy and Environmental Research Center (EERC).6,7,23,24 They monitored Hg vapor speciation at the SCR inlets and outlets at seven stations rated from 650 to 1300 MW. Four different SCR catalyst manufacturers were represented, although four sites used honeycombs from the same manufacturer. Two sites were monitored again after ⬃1 yr. The publications in open literature included the coal heating value, moisture, ash, sulfur, Cl content, and Hg content. Final reports to the U.S. Department of Energy and EPRI25,26 reported complete proximate and ultimate analyses of numerous samples taken during the testing campaigns. All of the coals were hv bituminous from mines in either the Midwestern or Eastern United States, except for the subbituminous burned at Site 1 and the subbituminous/bituminous blend burned at S8. However, the subbituminous at Site 1 generated an unusually high proportion of Hg(II) at the SCR inlet, because of the high UBC level from this cyclone furnace. The Cl contents of the hv bituminous coals are typical, except for the relatively low values in the hv bituminous coals from Sites 4 and 5. The testing sites were characterized with the furnace rating; firing configuration; in-furnace control technologies for NOX, SOX, and particulates; SCR catalyst vendor and type; space velocity; and age. Whereas NOX conversion efficiencies were not reported, the inlet and outlet NO levels and NH3 slip were. As expected, the NH3 slip levels were usually well below 1 ppm, so NOX conversion efficiencies were assigned from the NO concentrations across the SCR, and the NH3/NO ratios were set to 1870 Journal of the Air & Waste Management Association

the same values. The only omitted input data are the SCR temperature for Site 6 and the monolith pitch for Site 5. Model Validations Each test was simulated by entering the reported operating conditions into the computerized implementation of the SCR Hg0 oxidation mechanism. A preliminary calculation loop used the coal properties to assign a complete flue gas composition by varying the air flow rate to match the estimated flue gas O2 concentration to a reported value. The governing equations were then solved to determine the extents of NO reduction and Hg0 oxidation along the monolith. Rate parameters were adjusted to tune the predictions to the measured values for each catalyst type. Then, without additional parameter adjustments, simulations were run to cover the variations in HCl concentration, NH3/NO ratio, temperature, monolith dimension, and so forth. Hg transformations across the SCRs are characterized by the conversion efficiency for inlet Hg0, ␩Hg0, which is defined as follows: CHg0 ⫺ OUTCHg0 IN CHg0

IN

␩ Hg 0 ⫽

(6)

Because the total measured Hg vapor concentrations stayed the same across the SCRs, within experimental uncertainty, the definition of ␩Hg0 is not subject to variations in the total flue gas flow rate. The input operating conditions are collected in Table 1. The laboratory-scale tests with synthetic flue gas by Bock et al.22 provide the best coverage of temperature and monolith specifications; pilot-scale tests by Lee et al.27 cover the broadest range of HCl concentrations; and tests by Machalek et al.15 cover the broadest ranges of space velocity and NH3/NO. These three datasets were used to demonstrate that the proposed SCR Hg0 oxidation mechanism can depict the influences of coal quality, space velocity, NH3/NO ratio, temperature, and monolith specifications. Because the tests by Machalek et al.15 and Lee et al.27 were conducted at a single temperature, the proportion of Hg(II) at the SCR inlet was fixed for all of the test cases. However, over the temperature range examined by Bock et al.,22 the inlet Hg(II) percentage with 60 ppm HCl varied from 7% to 16% for temperatures from 280 °C to 420 °C, as monitored in runs without an active catalyst. This baseline was applied to all of the tests with 60 ppm HCl to evaluate ␩Hg0, assuming that homogeneous Hg0 oxidation was essentially independent of the catalytic oxidation. Whereas no such baseline was reported for the tests with 10 ppm HCl, the homogeneous oxidation rate was deemed to be negligible at the lower HCl level. Volume 55 December 2005

Niksa and Fujiwara Table 1. Input operating conditions for the SCR database. Catalyst Variables Laboratory-scale Lee et al.16 Bock et al.22 Pilot-scale Lee et al.27 Slipstream Machalek et al.15 Full-scale7,23–26 S1 S2–1 S2–2 S3 S4–1 S4–2 S5 S6 S8

T (ⴗC)

GHSV (hrⴚ1)

HCl (ppm)

NO (ppm)

NH3/NO

Typea

Pitch (mm)

Shapeb

350 280–420

2609 4800–5600

0–202 10, 60

350 400–530

0.9 1.0

H H, P

8.3 4.2–6.7

S S, R

Cormetech Argillon

342

2943

8–241

465–875

0.9

H

8.2

S

Cormetech

371

2500–7800

1.2

350

0.0–1.0

H

383 350 350 364 363 363 335

1800 2125 2125 3930 2275 2275 3700 3800 3100

900 — 415 370 730 600 280 330

0.90 0.95 0.95 0.90 0.91 0.91 0.75 0.85

H P P H H H P H H

336

4 91 42 60 24 15 28 79 49

10 9 5.6 5.6 7.4 8 8 7 7

Manufacturer

S S C C S S S T S S

Cormetech Argillon Argillon KWH Cormetech Cormetech Haldor Topsoe Cormetech Cormetech

Notes: H ⫽ honeycomb; P ⫽ plate; S ⫽ square channel; R ⫽ rectangular channel, C ⫽ circular channel; T ⫽ triangular channel.

Impact of SCR Operating Conditions. Simulations of the laboratory-scale data of Lee et al.16 for synthetic flue gas predicted essentially complete Hg0 oxidation whenever HCl was present and no Hg0 oxidation without it, in accord with the measured values. The evaluation with the laboratory data of Bock et al.22 on monolith specifications appears in the top panel of Figure 1. A single set of rate parameters was used to predict the impact of cell size on Hg0 oxidation in the square monolith channels, assuming that these extruded honeycombs had the same composition and preparation conditions. The SCR oxidation mechanism correctly predicts more Hg(II) for the more dense honeycomb and for progressively hotter temperatures with both honeycombs and a slightly stronger temperature dependence for the more open honeycomb. The predictions are within experimental uncertainty throughout, except for the hottest temperature with the more dense honeycomb. The evaluation for the plate catalyst in the lower panel is based on a lower activation energy for the Hg0 surface oxidation, which inverts the temperature dependence at both HCl concentrations. The mechanism correctly predicts less Hg0 oxidation and a stronger temperature dependence for the lower HCl concentration and less Hg0 oxidation for progressively hotter temperatures at both HCl levels. These predictions are within experimental uncertainty throughout. The evaluation with the pilot-scale data of Lee et al.27 appears in Table 2. The stringency of this evaluation may be eroded by the large breaches in the Hg balances, particularly with Turris coal, which were attributed to Hg adsorption on the catalyst. Notwithstanding, this dataset Volume 55 December 2005

Figure 1. Evaluation with laboratory data from Bock et al.22 on (top) honeycomb and (bottom) plate monoliths for various temperatures. Journal of the Air & Waste Management Association 1871

Niksa and Fujiwara Table 2. Evaluation with pilot-scale data from Lee et al.27 Coal Galatia Turris Crown II Black Thunder Black Thunder

Hg Closure

Measured ␩Hg0

Predicted ␩Hg0

72.7 47.8 74.6 70.8 90.2

95.6 93.6 89.1 35.7 28.6

95.4 90.9 89.9 32.5 32.5

was simulated with a single set of rate parameters, because the same catalyst was used with the four coal samples. The predictions for all three of the bituminous coals are ⱖ90%, in accord with the data, and correctly identify the much lower extent of Hg0 oxidation for the subbituminous coal. All of the predictions for this dataset are within experimental uncertainties. This performance is significant, because the Hg0 oxidation behavior for the very broad range of HCl concentrations generated by these diverse coals (compare with Table 1) was accurately predicted with a single set of rate parameters. The inhibition of Hg0 oxidation by NH3 injection over the full range of NH3/NO is apparent in the evaluation in Figure 2. These data were recorded in the slipstream tests by Machalek et al.15 with subbituminous coal. When the NH3/NO ratio was increased from zero to unity, NO reduction efficiencies increased in direct proportion to the NH3/NO ratio, except for the highest NH3 injection rate. The predictions are nearly exact except for the slight underprediction for the highest injections. The measured extents of Hg0 oxidation diminish except for NH3/NO ratios ⬎0.8. The predictions accurately depict the inhibition of Hg0 oxidation by NH3 over the full range, except that they do not exhibit the saturation at the highest NH3

Figure 2. Evaluation with slipstream data from Machalek et al.15 with subbituminous-derived flue gas on a honeycomb monolith for various NH3/NO ratios at 371 °C for a GHSV of 3000 hr⫺1. 1872 Journal of the Air & Waste Management Association

Figure 3. Evaluation with subbituminous-derived flue gas on a honeycomb monolith for various GHSVs at 371 °C with (F and lower curve) and without (E and upper curve) NH3.15

injections. Such saturation is hard to rationalize, because the NH3 surface coverage should continue to increase for higher NH3/NO ratios, especially for NH3/NO ratios over unity, because none of the trailing edge of the monolith would remain free of NH3 in the superstoichiometric cases, as it is in the substoichiometric runs. Notwithstanding this apparent discrepancy, the mechanism accurately describes NH3 inhibition with a single set of rate parameters. The same rate parameters yielded the accurate predictions for space velocities from 3000 to 7800 hr⫺1 in Figure 3, which depicts the dependence on space velocity for two NH3/NO ratios. Whereas the predictions for NH3/NO at 0.8 are within experimental uncertainty over the full range of space velocity, those for no NH3 injection are accurate up to 5000 hr⫺1 but not for the higher velocities. Space velocities ⬎4000 hr⫺1 are not found in utility SCR applications. The evaluations with full-scale test data are subject to greater uncertainties because of the inherent variability of fuel properties and operating conditions in full-scale gas cleaning systems. Data taken from the four commonsource SCRs were first simulated with the same rate parameters assigned for pilot scale tests by Lee et al.27 with the same source material. Because of slight overpredictions, the rate constant for Hg0 oxidation had to be reduced, consistent with the expected loss of reactivity after long-term exposure to flue gas in the full-scale SCRs. Yet the four common-source full-scale monoliths were simulated with the same rate parameters. Only the fuel properties and SCR operating conditions were changed to match the values imposed during the tests. The tests with the KWH honeycomb were simulated with the same rate parameters, except that kHg was reduced by a factor of 5. Because of the distinctive shape of Halder Topsoe plate Volume 55 December 2005

Niksa and Fujiwara Table 3. Evaluation of Hg0 conversion efficiencies across full-scale SCRs.7,23-26 ␩Hg0 (%)

Inlet (␮g/Nm3)

Outlet (␮g/Nm3)

Site

Hg0

HgCl2

Hg0

HgCl2

Measured

Predicted

S1 S2–1 S2–2 S3 S4–1 S4–2 S5 S6 S8

6.0 7.2 5.5 16.6 12.8 8.2 7.6 3.7 54.9%

0.4 6.9 6.8 20.7 1.8 4.2 6.3 6.0 45.1%

5.0 1.4 1.5 11.6 2.2 4.1 2.6 1.5 7.2%

1.8 15.8 11.2 22.7 10.2 7.2 11.9 7.3 92.8%

16.7 80.6 72.7 30.1 82.8 50.0 65.8 59.5 86.9

19.9 85.5 64.9 28.9 61.4 48.0 68.3 78.9 77.7

monoliths, they were simulated as a honeycomb monolith with triangular channels with the same parameters as the same-source honeycomb, except that kHg was reduced by a factor of 2. Similarly, simulations for the plate catalysts were based on the parameters assigned for the tests by Bock et al.22 in this configuration, except that kNO had to be increased to match the reported NH3 slip, and kHg had to be reduced. The evaluation of full-scale SCRs appears in Table 3. Note that all of the measured Hg species concentrations data are reported in micrograms per normal meters cubed except for Site 8, where the speciation is expressed as a percentage of the total Hg concentration. The final two columns compare the measured and predicted Hg0 oxidation efficiencies. The agreement is within experimental uncertainty, except for Site 6 and, perhaps, for Site 4 –1. The agreement is especially significant for the four cases with the common-source monoliths, because none of the rate parameters were adjusted to depict the impact of substantial variations in coal quality and SCR operating conditions. Moreover, the predicted NH3 slip was always well below 2 ppm, in accord with the regulatory standard. Clearly, the proposed SCR Hg0 oxidation mechanism accurately interprets test data from fullscale SCRs with reasonable parameter assignments. Although more data evaluations are needed to draw definitive conclusions, the early indications are that our SCR Hg0 oxidation mechanism accurately represents how coal quality affects Hg0 oxidation, given test data on one coal to tune in the rate parameters for the catalyst under consideration. The evaluation in Table 3 is especially pertinent, because the HCl concentrations in this database range from 4 to 91 ppm, which covers most of the typical range in utility gas cleaning systems. Parametric Sensitivity. Based on the satisfactory performance of the mechanism in interpreting the laboratoryscale, pilot-scale, and full-scale databases, the mechanism Volume 55 December 2005

Figure 4. Influences of coal-Cl (solid curve, lower axis) and monolith pitch (dashed curve, upper axis) on ␩Hg0.

was used to illustrate the impact of the important SCR operating conditions on ␩Hg0. The baseline conditions were specified from Site 3 in the EERC full-scale database for which there was 59 ppm HCl in the flue gas and NH3/NO at 0.90, T at 364 °C, and GHSV at 3930 hr⫺1. The monolith was a square honeycomb with 7.4-mm pitch. The impacts of variations in coal-Cl and monolith pitch appear in Figure 4. Coal-Cl is expressed in terms of the associated HCl concentration in flue gas derived from a typical hv bituminous coal from the Eastern United States. The Hg0 oxidation efficiency increases for progressively higher HCl levels, in direct proportion to HCl below ⬃40 ppm. As seen in our mechanisms for in-flight Hg0 oxidation2 and for Hg2⫹ retention in wet FGDs,28 the effect diminishes for the range of HCl concentrations with typical hv bituminous coals. Hence, most of the variation in ␩Hg0 in the EERC database should be attributed to variable SCR operation rather than to coal-Cl variations, because almost all of the samples were hv bituminous coals. Only Sites 1, 4 –1, 4 –2, and 5 had HCl concentrations ⬍40 ppm. Figure 4 also shows that monoliths with smaller pitches, all else the same, are more effective Hg0 oxidation reactors. The direct reason is that film transport rates accelerate for progressively smaller monolith channels, so the overall Hg0 oxidation is also accelerated. However, this same effect also applies to NH3, which inhibits Hg0 oxidation. Because faster NH3 transport rates also accelerate the NO reduction rate across the inlet region, the downstream region, where there is little NH3 to inhibit Hg0 oxidation, becomes longer with smaller channels and sustains more Hg0 oxidation. In our calculations, the surface coverage of HCl increases by ⱖ2 orders of magnitude Journal of the Air & Waste Management Association 1873

Niksa and Fujiwara

Figure 5. Influences of GHSV (solid curve, lower axis) and SCR temperature (dashed curve, upper axis) on ␩Hg0.

from the upstream to the downstream regions of the SCR monolith, so both effects are important. The impact of greater NH3/NO ratios is similar to pitch. The computed values of ␩Hg0 for ratios from 0.65 to unity are nearly constant at 29% below 0.85, then decrease to 19% at a ratio of unity. With higher NH3/NO ratios, surface NH3 concentrations are higher, so that surface HCl concentrations remain lower; in other words, the length of the inlet region of strong NH3 inhibition on Hg0 oxidation becomes longer for progressively greater NH3/NO ratios. The influences of GHSV and SCR temperature appear in Figure 5. Extents of Hg0 oxidation fall for progressively greater GHSV simply because there is less time available for chemistry on the monolith. This is a strong effect for the range of GHSV in the full-scale database, so GHSV variations should be recognized as significant sources of the measured variations in ␩Hg0. Extents of Hg0 oxidation slightly increase for progressively hotter SCR temperatures, consistent with the weak temperature dependence in the laboratory-scale data for honeycomb monoliths in Figure 1. Bear in mind that the predicted temperature dependence for plate monoliths was inverted from the dependence for honeycombs, in accord with data from Bock et al.22 More data are needed to determine whether the forms of such temperature dependences are really determined by the monolith configuration or the catalyst composition or by some combination of factors. For the time being, the predicted temperature dependence in Figure 5 should not be regarded as a general characteristic of Hg0 oxidation on SCR monoliths. CONCLUSIONS Previous simulation studies of catalytic NO reduction established that NH3 strongly adsorbs onto acidic sites on SCR catalysts of V2O5 on TiO2, but NO reacts from the gas 1874 Journal of the Air & Waste Management Association

phase or as a weakly adsorbed species. An Eley-Rideal rate expression, in conjunction with film transport coefficients for developing laminar boundary layers, accurately predicts the NO reduction performance of full-scale SCRs in both honeycomb and plate configurations. According to the proposed SCR Hg0 oxidation mechanism, a simple premise connects this capability to the Hg0 oxidation behavior on SCRs: HCl competes for surface sites with NH3, and Hg0 contacts these chlorinated sites either from the gas phase or as a weakly adsorbed species. Accordingly, Eley-Rideal rate expressions have been proposed for both NO reduction and Hg0 oxidation which, in combination with the transport coefficients that have already been validated for NO reduction, describe the Hg0 oxidation rate as a function of the NH3/NO ratio and concentrations of Hg0 and HCl at the monolith inlet, the monolith pitch and channel shape, and the SCR temperature and GHSV. This mechanism explicitly accounts for the inhibition of Hg0 oxidation by NH3, which is essential, because NH3, as the substoichiometric reductant, necessarily vanishes along the monolith from its inlet concentration. The monolith thereby sustains two chemically distinct regions. In the inlet region, strong NH3 adsorption minimizes the coverage of chlorinated surface sites, so NO reduction inhibits Hg0 oxidation. But once the NH3 has been consumed, the chlorinated surface coverage surges by at least 2 orders of magnitude, and the Hg0 oxidation rate rapidly accelerates, even while the HCl concentration in the gas phase is uniform. Hence, the various factors affecting Hg0 oxidation can be understood in terms of how they affect the extents of these two regions. Factors that accelerate the transport rates of reactants onto the catalyst walls, such as smaller channel pitches and converting from square to circular channels, shorten the length of the NO reduction region and thereby promote Hg0 oxidation. Factors that enhance surface chlorination, such as higher inlet HCl concentrations and lower NH3/NO ratios, also promote Hg0 oxidation. Evaluations of the predictions with laboratory-scale and pilot-scale SCRs and slipstream test data demonstrated that the mechanism accurately depicts the impact of variations in coal-Cl, NH3/NO, GHSV, temperature, monolith pitch, and channel shape. Without any heuristic adjustments to the rate parameters, this mechanism accurately interpreted the reported tendencies for greater extents of Hg0 oxidation on honeycomb monoliths with smaller channel pitches and hotter temperatures. By adjusting the activation energy in the surface Hg0 oxidation rate, the analysis also interpreted the tendency for lower extents of Hg0 oxidation for hotter temperatures on plate monoliths within experimental uncertainties. With a single set of rate parameters, the mechanism also described the inhibition of Hg0 oxidation by NH3 Volume 55 December 2005

Niksa and Fujiwara for NH3/NO ratios from zero to 0.9 and correctly predicted that higher NH3/NO ratios inhibit Hg0 oxidation over a broad range of GHSV. Perhaps most important for practical applications, the mechanism with a single set of rate parameters reproduced the reported extents of Hg0 oxidation on a single catalyst for four coals that generated HCl concentrations from 8 to 241 ppm, which cover the entire range encountered in the U.S. utility industry. A similar performance was also demonstrated for full-scale SCRs, with only slightly lower Hg0 oxidation reactivities, consistent with the expected loss of reactivity after long-term exposure to flue gas in the full-scale SCRs. ACKNOWLEDGMENTS The authors acknowledge the assistance of George Offen and Paul Chu of EPRI in obtaining detailed monolith specifications for the full-scale validation studies. This study was sponsored in part by the New Energy Development Organization Clean Coal Technology Center under the Toxic Metals Project, and also by the Coal Research Laboratory, Idemitsu Kosan Company, Ltd. REFERENCES 1. Niksa, S.; Fujiwara, N.; Fujita, N.; Tomura, K;. Moritomi, H.; Tuji, T.; Takasu, S. A Mechanism for Hg Oxidation in Coal-Derived Exhausts; J. Air & Waste Manage. Assoc. 2002, 52, 894-901. 2. Niksa S.; Fujiwara, N. Predicting Extents of Mercury Oxidation in CoalDerived Flue Gas; J. Air & Waste Manage. Assoc., 2005, 55, 930–939. 3. Nakayama, Y.; Ohishi, T.; Nakamura, S.; Okino, S.; Honjo, S. Development of Mercury Oxidizing Catalyst. In Proceedings of the U.S. Environmental Protection Agency-Department of Energy-EPRI Combined Power Plant Air Pollutant Control Symposium: The MEGA Symposium; U.S. Environmental Protection Agency-Department of Energy-EPRI: Washington, DC, 2003. Paper No 179. 4. Blythe, G.; Richardson, C., Strohfus, M.; Lee, A.; Rhudy, R.; Lani, B. Pilot Testing of Oxidation Catalysts for Enhanced Mercury Control by Wet FGD. In Proceedings of the U.S. Environmental Protection AgencyDepartment of Energy-EPRI Combined Power Plant Air Pollutant Control Symposium: The MEGA Symposium; U.S. Environmental Protection Agency-Department of Energy-EPRI: Washington, DC, 2004. 5. Richardson, C.; Machalek, T.; Miller, S. Effect of NOx Control Processes on Mercury Speciation in Utility Flue Gas; J. Air & Waste Manage. Assoc. 2002, 52, 941-947. 6. Laudal, D.L.; Thompson, J.S.; Pavlish, J.H.; Brickett, L.; Chu, P.; Srivastava, R.K.; Lee, C.W.; Kilgroe, J. Mercury Speciation at Power Plants Using SCR and SNCR Control Technologies; Environ. Manager 2003, 16-22. 7. Chu, P.; Laudal, D.; Brickett, L.; Lee, C.W. Power Plant Evaluation of the Effect of SCR Technology on Mercury. In Proceedings of the U.S. Environmental Protection Agency-Department of Energy-EPRI Combined Power Plant Air Pollutant Control Symposium: The MEGA Symposium; U.S. Environmental Protection Agency-Department of Energy-EPRI: Washington, DC, 2003. Paper No. 106. 8. Lietti, L.; Ramis, G.; Berti, F.; Toledo, G.; Robba, D.; Busca, G.; Forzatti, P. Chemical, Structural, and Mechanistic Aspects on NOX SCR over Commercial and Model Oxide Catalysts; Catalysis Today 1998, 42, 101-116. 9. Svachula, J.; Ferlazzo, N.; Forzatti, P.; Tronconi, E. Selective Redction of NOx by NH3 over Honeycomb DeNOXing Catalysts; Ind. Eng. Chem. Res. 1993, 32, 1053-1060. 10. Willi, R.; Roduit, B.; Koeppel, R.A.; Wokaun, A.; Baiker, A. Selective Reduction of NO by NH3 over Vanadia-Based Commercial Catalyst: Parametric Sensitivity and Kinetic Modeling; Chem. Eng. Sci. 1996, 51, 2897-2902. 11. Tronconi, E.; Forzatti, P.; Gomez Martin, J.P.; Malloggi, S. Selective Catalytic Removal of NOx: A Mathematical Model for Design of Catalyst and Reactor; Chem. Eng. Sci. 1992a, 47, 2401-2406. 12. Tronconi, E.; Forzatti, P. Adaquacy of Lumped Parameter Models for SCR Reactors with Monolith Structure; AIChE J. 1992b, 38, 201-210.

Volume 55 December 2005

13. Giudici, R.; Tronconi, E. Laminar Flow and Forced Convection Heat Transfer in Plate-Type Monolith Structures by a Finite Element Solution; Int. J. Heat Mass Transfer 1996, 39, 1963-1978. 14. Beretta, A.; Orsenigo, C.; Forzatti, P.; Tronconi, E.; Forzatti, P.; Berti, F. Analysis of the Performance of Plate-Type Monolithic Catalysts for scr DeNOx Applications; Ind. Eng. Chem. Res. 1998, 37, 2623-2633. 15. Machalek, T.; Ramavajjala, M.; Richardson, M.; Richardson, C. Pilot Evaluation of Flue Gas Mercury Reactions across an SCR Unit. In Proceedings of the U.S. Environmental Protection Agency-Department of Energy-EPRI Combined Power Plant Air Pollutant Control Symposium: The MEGA Symposium; U.S. Environmental Protection Agency-Department of Energy-EPRI: 2003, Washington, DC. Paper No. 64. 16. Lee, C.W.; Srivastava, R.K.; Ghorishi, S.B.; Hastings T.W.; Stevens, F.M. Study of Speciation of Mercury under Simulated SCR NOX Emission Control Conditions. In Proceedings of the U.S. Environmental Protection Agency-Department of Energy-EPRI Combined Power Plant Air Pollutant Control Symposium: The MEGA Symposium; U.S. Environmental Protection Agency-Department of EnergyEPRI: 2003, Washington, DC. Paper No. 41. 17. Spitznogle, G.; McDonald, K.; Lin, C.; Vesanen, A.; Toole, A.; Duellman, D. Oxidation of Mercury across a Slipstream Reactor Equipped with Various Catalyst Formulations. In Proceedings of the 8th Electric Utilities Environmental Conference; Tucson, AZ, 2005. Paper No. A96. 18. Forzatti, P. Present Status And Perspectives in De-NOX SCR Catalysis; Appl. Catalysis A: Gen. 2001, 222, 221-236. 19. Koebel, M.; Elsener, M. SCR of NO Over Commercial DeNOx Catalysts: Experimental Determination of Kinetic and Thermodynamic Parameters; Chem. Eng. Sci. 1998, 53, 657-669. 20. Tronconi, E.; Beretta, A.; Elmi, A.S.; Forzatti, P.; Malloggi, S.; Baldacci, A. A Complete Model of SCR Monolith Reactors for the Analysis of Interacting NOx Reduction and SO2 Oxidation Reactions; Chem. Eng. Sci. 1994, 439, 4277-4287. 21. Bai, H.; Chwu, J-W. Theoretical Analysis of SCR Catalysts; J Environ. Eng. 1997, 123, 431-36. 22. Bock, J.; Hocquel, M.J.T.; Unterberger, S.; Hein, K.R.G. Mercury Oxidation across SCR Catalysts of Flue Gas with Varying HCl Concentrations. In Proceedings of the U.S. Environmental Protection Agency-Department of Energy-EPRI Combined Power Plant Air Pollutant Control Symposium: The MEGA Symposium; U.S. Environmental Protection Agency-Department of Energy-EPRI: 2003, Washington, DC. Paper No. 233. 23. Laudal, D.L.; Thompson, J.S.; Pavlish, J.H.; Brickett, L.; Chu, P.; Srivastava, R.K.; Lee, C.W.; Kilgroe, J. Evaluation of Mercury Speciation at Power Plants Using SCR and SNCR Control Technologies; Air Quality III; University of North Dakota Energy and Environmental Research Center: Washington, DC, 2001. 24. Laudal, D.L.; Wocken, C.A.; Chu, P.; Brickett, L.; Lee, C.W. Evaluation of the Effect of SCR on Mercury Speciation and Emissions; Air Quality III; University of North Dakota Energy and Environmental Research Center: Washington, DC, 2001. 25. Laudal, D. L. Selective Catalytic Reduction Mercury Field Sampling Project; EPRI Report No. 1005400; EPRI: Palo Alto, CA, 2002. 26. Laudal, D. L. Effect of Selective Catalytic Reduction on Mercury, 2002 Field Studies Update; Final Report to U.S. Department of Energy under contract DE-FC26 –98FT40321; University of North Dakota Energy and Environmental Research Center: Washington, DC, 2003. 27. Lee, C.W.; Srivastava, R.K.; Ghorishi, S.B.; Karwowski, J.; Hastings T.W.; Hirschi, J. Effect of SCR Catalysts on Mercury Speciation. In Proceedings of the U.S. Environmental Protection Agency-Department of Energy-EPRI Combined Power Plant Air Pollutant Control Symposium: The MEGA Symposium; U.S. Environmental Protection Agency-Department of Energy-EPRI: 2004, Washington, DC. 28. Niksa S.; Fujiwara, N. The Impact of Wet FGD Scrubbing on Hg Emissions from Coal-Fired Power Stations; J. Air & Waste Manage. Assoc., 2005, 55, 970 –977.

About the Authors Stephen Niksa is an international consultant at Niksa Energy Associates. Naoki Fujiwara is a research engineer at the Coal Research Laboratories at Idemitsu Kosan, Co, Ltd. Address correspondence to: Stephen Niksa, Niksa Energy Associates, 1745 Terrace Drive, Belmont, CA 94002; fax: (650) 654-3179; e-mail: [email protected].

Journal of the Air & Waste Management Association 1875