2618 Ind. Eng. Chem. Res.

2001, 40, 2618-2627

Use of Simulation To Optimize NOx Abatement by Absorption and

Selective Catalytic Reduction
Andrew J. Sweeney and Y. A. Liu*
Honeywell Center of Excellence in Computer-Aided Design, Department of Chemical Engineering,
Virginia Polytechnic Institute and State University, Blacksburg, Virginia 24061

This paper describes the development of an ASPEN PLUS simulation model for a commercial
NOx abatement system involving both absorption and selective catalytic reduction (SCR). The
model helps identify operator guidelines and retrofit options needed to enable the commercial
system to operate efficiently during surges in NOx-laden fumes without incurring costly fines.
The resulting model applies a reactive-distillation module with a practical reaction set for NOx
absorption and implements a kinetic model for SCR. The simulation results agree well with
both design specifications and literature data and provide practical insights for optimum
operation and economical retrofits of the commercial system.

1. Introduction suggests retrofit options for RFAAP that can increase

the operating efficiency. Importantly, its success makes
The system under analysis involves two key pro- it an effective starting point for modeling similar
cesses: absorption and selective catalytic reduction situations.
(SCR). The absorption of nitrogen oxide (NOx) gases,
particularly NO2, into aqueous solutions is important
in both nitric acid production and the abatement of NOx- 2. NOx Abatement Process at RFAAP
laden effluent streams for environmental protection.1 The RFAAP process receives NOx-rich fumes from the
Downstream, SCR removes NOx by injecting ammonia nitrocellulose production line on site. The abatement
into the flue gas, where chemical reactions in the system combines an absorber tower to remove NO2 and
presence of a catalyst convert NO to nitrogen and an SCR catalyst vessel to remove NO.
water.2 Specifically, the absorber removes most of NO2 in the
Previous studies have investigated NOx absorption NOx fumes and converts a portion to nitric acid. The
and SCR individually using accurate but highly complex remaining NOx fumes proceed to a catalyst vessel, where
mass-transfer and reaction-kinetics calculations. They ammonia reduces NOx to atmospheric nitrogen, N2.
have not, however, thoroughly treated these two pro- From the catalyst vessel, the gases go to the stack and
cesses in combination and, consequently, little literature out to the atmosphere. Figure 1 shows the process, with
exists on the subtleties of linking these two distinct, stream values drawn from the design specifications.
emerging technologies. Yet, more and more chemical The absorber houses 16 bubble-cap trays in the upper
and environmental engineers in the field have begun portion of the tower and 2 spray-scrubber sections in
exploiting opportunities to effectively couple the pro- the bottom. The NOx-laden fumes enter the bottom of
cesses to maximize NOx removal, making the need for the tower and contact with a mist of weak nitric acid.
more research critical. Filtered water enters at the top and flows down as the
Such a system is, in fact, already in place at the fumes flow up. The fumes leave the top of the column
Radford Facility and Army Ammunition Plant (RFAAP), and enter a demister tank, where they impinge on
Radford, VA. In response to the knowledge gap in the stainless steel mesh screens, removing the entrained
literature, this work simulates the multimillion-dollar liquid that could damage the catalyst vessel.
RFAAP NOx abatement system to develop operator The SCR reaction occurs at 316 °C. The fumes leave
guidelines and low-cost retrofit options that will ensure the absorber at 27 °C and are heated in three stages:
efficient operation during surges in NOx-laden fumes (1) a steam preheater heats the vapor stream to 38 °C;
that currently produce higher-than-allowed NOx out- (2) a heat exchanger, called an economizer, uses pos-
flows. This effort will avoid frequent shutdowns and treaction gases to heat the stream to approximately 177
costly fines of up to $25 000/day from the Virginia °C; and (3) a direct-fired heater uses natural gas
Department of Environmental Quality (VDEQ) for NOx combustion to increase the temperature to 316 °C.
outflows above the hourly average limit of 125 ppm by
volume (ppmv).
Combining real plant data with literature parameters, 3. ASPEN PLUS
we have developed an ASPEN PLUS model that simu- ASPEN PLUS is currently the most popular chemical
lates system performance, and the results compare process simulation software used. Given its wide use,
favorably with both plant data and literature for we choose it for this process to determine if it could
processes with similar conditions. The simulation pre- simulate the complexities of NOx absorption and SCR.
dicts system sensitivity to various process variables and Several factors present a challenge when considering
the viability of the simulation software in this process.
* To whom correspondence should be addressed. Phone: For example, what are the chemical reactions important
(540) 231-7800. Fax: (540) 231-5022. E-mail: [email protected]. to NOx absorption and how are those reactions imple-
10.1021/ie0005295 CCC: $20.00 © 2001 American Chemical Society
Published on Web 04/26/2001
 Ind. Eng. Chem. Res., Vol. 40, No. 12, 2001 2619

Figure 1. Block flow diagram of the NOx abatement system provided by RFAAP.

Table 1. Full Reaction and Absorption Mechanism for NOx and Water3,a
N2O4 pathway N2O3 pathway NO2 pathway
Gas Phase
2NO + O2 T 2NO2 (1) NO + NO2 T N2O3 (6) 3NO2 + H2O T 2HNO3 + NO (10)
2NO2 T N2O4 (2)
Transport
N2O4(g) T N2O4(l) (3) N2O3(g) T N2O3(l) (7) NO2(g) T NO2(l) (11)
HNO3(g) T HNO3(l) (12)
Liquid Phase
N2O4 + H2O T HNO3 + HNO2 (4) N2O3 + H2O T 2HNO2 (8) 2NO2 + H2O T HNO3 + HNO2 (13)
3HNO2 T HNO3 + H2O + 2NO(g) (5) 2HNO2 + O2 T 2HNO3 (9)
a (l) refers to a liquid-phase component, and (g) refers to a gas-phase component.
mented in the software? How do we deal with the absorber. The high absorption capacity of NOx stems
electrolyte species in the liquid phase inside the absorp- from the high thermodynamic stability of aqueous nitric
tion column? How do we handle combined reactions and acid, the final product of NOx absorption. This stability,
separation in an absorption column? We shall address in turn, results from the dissociation of the acid to the
these and other fundamental and practical issues below. nonvolatile nitrate ion, NO3-. As a strong acid, aqueous
nitric acid exists mostly as NO3-.
4. Simulating NOx Absorption: Practical ASPEN PLUS’s electrolyte, nonrandom-two-liquid
Reaction Set (NRTL) thermodynamic model includes electrolyte ac-
tions in the absorber’s aqueous environment. This model
The conditions for treating effluents from combustion requires all electrolytic reactions to be specified; for our
sources generally preclude using absorption but are problem, the key reaction is
ideal for SCR. However, SCR does not deal with NO2
as efficiently as NO. Therefore, cases such as RFAAP
require both processes. Importantly, unlike most gas HNO3 + H2O f NO3- + H3O+ (14)
absorption, NOx absorption is driven primarily by
chemical reaction, while the rate-limiting step is inter- Although the reaction is labeled “global” in the
facial mass transfer. Table 1 lists the most general simulation (occurring in all units), it must also appear
reactions, (1)-(13), for the process according to Miller.3 as a reactive-distillation reaction in the absorber stages.
Because the absorber promotes reaction as well as Otherwise, ASPEN PLUS assumes that the reaction
mass transfer, ASPEN PLUS’s reactive-distillation does not occur in the absorber.
model describes it mathematically. Applying the assumptions described in the following
In addition to reactions (1)-(13) listed in Table 1, we sections to the reactions listed in Table 1 produces a
also need to consider the action of electrolytes in the practical model for the NOx absorption at RFAAP. These
2620 Ind. Eng. Chem. Res., Vol. 40, No. 12, 2001

following reaction, (4-5), at equilibrium in the liquid

phase:

3N2O4 + 2H2O f 4HNO3 + 2NO(g) (4-5)

4.4. Assumption IV: Elimination of HNO2. Sev-

eral factors support elimination of HNO2. First, com-
bining reactions (4) and (5) (assumption III) to eliminate
the intermediate HNO2 according to Miller3 implies that
the concentration of HNO2 remains very low and
relatively constant because reaction (5) consumes it as
soon as reaction (4) produces it. HNO2 does not exist
for extended periods because it is a very unstable
compound. Second, literature data for the liquid-phase
oxidation of HNO2 [reaction (9): 2HNO2 + O2 ) 2HNO3]
remain unavailable,3 and the liquid-phase concentration
Figure 2. Transient oxidation of NO via reaction (1) for various of O2 is very low. Therefore, we can neglect the effect of
initial partial pressures. HNO2 oxidation in the liquid-phase reactions, (8) and
(9).
assumptions stem from the relative importance of 4.5. Assumption V: Treatment of Reaction (6) as
certain reactions and components, and the simulation’s Being in an Instantaneous Equilibrium. The as-
realistic performance (sections 7 and 8) validates them. sociation of NO and NO2 molecules to produce N2O3
4.1. Assumption I: Inclusion of Reaction (1) as [reaction (6)] is much slower than reaction (2), equili-
a Kinetic Rate Equation. Reaction (1), the oxidation brating in 0.1 s [versus 0.0001 s for reaction (2)].5 In
of NO to NO2, remains important in NOx chemistry. In addition, N2O3 is much less stable than N2O4 according
cases where NO represents the major contributor to the to Miller.3 The assumption of equilibrium levels of N2O4
total NOx, it has been identified as a major rate and N2O3 gives an appreciable amount of N2O4 but very
limitation in the overall absorption mechanism accord- little N2O3. At the absorber inlet, there is approximately
ing to Miller.3 In other words, reaction (1) is critical 100 times more N2O4 than N2O3 (4.8 × 10-2 kg‚mol/h
when producing NO2 from NO oxidation. Such is the N2O4 versus 4.1 × 10-4 kg‚mol/h N2O3).
case in nitric acid production, where ammonia oxidizes Some researchers (e.g., Thomas and Vanderschuren6)
to NO under high pressure, which in turn oxidizes to claim that the N2O3 pathway dominates, but this applies
NO2. However, at RFAAP, NOx fed to the system only in systems with higher pressures and higher NO-
consists mostly of NO2, making the reaction less critical. to-NO2 ratios. In these cases, N2O3 can outstrip N2O4
Because the oxidation of NO is irreversible according and also certainly absorbs faster than NO or NO2. Given
to Miller,3 we treat reaction (1) as a forward reaction that N2O4 exists in higher concentrations and thus will
only. Using reaction-rate data presented by Suchak and absorb and react at a faster rate, N2O3 absorption
Joshi,4 we calculate the time-dependent change of the represents a secondary mechanism in NOx absorption.
fraction of entering NO oxidized by reaction (1) (preex- In the case of RFAAP, we can neglect the N2O3 pathway
ponential ) 1.814 × 10-3 kPa-1 and activation energy with minimal effect on the simulation accuracy. How-
) 1.658 kcal/g‚mol). Figure 2 shows the results at ever, we will include the production and absorption of
varying NO partial pressures for the starting conditions N2O3 for the sake of completeness.
of the fume feed to the bottom of the column. With a If we combine reactions (8) and (5) to eliminate
gas flow rate of 113.3 CMM (4000 standard cubic feet transient HNO2 as we did with reaction (4-5) and
per minute, SCFM) through the column, the total void multiply the reaction by 2, the result becomes
volume allows a residence time of 15 s. According to
Figure 2, less than 15% of the NO oxidizes to NO2. This 3N2O3 + H2O f 2HNO3 + 4NO(g) (8-5)
result, together with the fact that NO2 represents the
major component of the total NOx fed to the absorber, We rewrite this reaction as
justifies elimination of reaction (1) in similar cases.
However, for completeness, we include reaction (1) in 6NO + 6NO2 + 2H2O f 4HNO3 + 8NO(g)
the model.
4.2. Assumption II: Treatment of Reaction (2) where 3NO cancels on each side of the equation, leaving
as Being in an Instantaneous Equilibrium. Reac- us with reaction (4-5) with 6NO2 replaced with 3N2O4.
tion (2), 2NO2 ) N2O4, proceeds quickly and reversibly 4.6. Assumption VI: Neglect of the Vapor-Phase
and equilibrates rapidly according to Miller3 and Matasa Acid Production. Because researchers typically treat
and Tonca.5 Because it typically achieves equilibrium the vapor-phase production of nitric and nitrous acids
in 0.0001 s,5 it is effectively in equilibrium instanta- as negligible,3 we can neglect reaction (10). Note that
neously with respect to other time-dependent concen- the mass-transfer reaction (12) for HNO3 will remain.
trations. Therefore, although the two compounds cannot 4.7. Assumption VII: Treatment of Reaction (13).
be isolated, we can always determine the concentration NO2 has a lower solubility and reactivity than N2O4.
of NO2 and N2O4 from the total NO2 + N2O4. In this Suchak and Joshi4 state that absorption of NO2 as such
work, NO2* denotes NO2 + 2N2O4. is negligible. Similarly, Miller3 says that the NO2 routes
4.3. Assumption III: Combination of Reactions are slow relative to those involving the reactants N2O4
(4) and (5). Reactions (4) and (5) are fast and equili- and N2O3 and can usually be neglected in evaluating
brate rapidly in the liquid film, especially for tray the nitric acid absorption performance. Newman and
absorbers as is the case at RFAAP. Miller3 and others Carta7 state that “the absorption of NO2 in water has
suggest representing them in combination, yielding the been studied extensively. All studies indicate that at
 Ind. Eng. Chem. Res., Vol. 40, No. 12, 2001 2621

Table 2. Parameter Values for the Reduction of NO and

NO2 with Ammonia
symbol units value source
A1 kg‚mol/(s‚m3) 15 × 106 a
E1 kcal/g‚mol 15 b
a unitless 1 b
b unitless 0.5 b
c unitless 5 × 10-4 d
pNO kPa c
pO2 kPa c
pNH3 kPa c
A2 kg‚mol/(s‚m3) 15 × 106 e
Figure 3. Illustration of the equilibrium conditions for the three E2 kcal/g‚mol 14.5 e
NOx pathways. x unitless 1 e
y unitless 0.5 e
z unitless 5 × 10-4 e
pNO2 kPa c
A3 kg‚mol/(s‚m3) 15 × 106 e
E3 kcal/g‚mol 10 e
m unitless 1 e
n unitless 5 × 10-4 e
A4 kg‚mol/(s‚m3) 0.1 e
E4 kcal/g‚mol 10 e
d unitless 0 e
e unitless 0.3 e
pH2O kPa c
pN2 kPa c
a Willi et al.9 b Marangozis.10 c Denotes that the value varies

depending on feed and catalyst-vessel operating conditions. d Most

research suggests zero order with respect to NH3; however, the
minimal value shown here ensures the cessation of reaction upon
exhaustion of the NH3 supply. e These values for equations (19)-
(21) were assumed based on the data for eq 18.
Figure 4. Final reaction and absorption mechanism after the
application of assumptions I-VII. (4-5) back to their reactants. Because reaction (4-5)
is in equilibrium and NO2 and N2O3 are active in the
partial pressures of NO2 greater than about 0.2 kPa
system, we also consider these NOx components in the
(0.002 atm) the absorption...occurs via N2O4.” The
reaction mechanism shown in Figure 4.
partial pressure of NO2 in the fume feed at RFAAP is
0.5 kPa (0.005 atm). If we combine reactions (13) and
(5) to eliminate transient HNO2 and multiply the 5. Selective Catalytic Reduction
reaction by 2, we obtain reaction (13-5). We see that We now turn to the catalyst vessel unit. In the
presence of a saturating level of oxygen, reactions (15)-
6NO2 + 2H2O f 4HNO3 + 2NO(g) (13-5) (17) represent the mechanism in the SCR.
this reaction is simply reaction (4-5) after replacing 4NH3 + 4NO + O2 f 4N2 + 6H2O (15)
6NO2 with 3N2O4.
4.8. Final Reaction Set. Figure 3 illustrates the 2NO2 + 4NH3 + O2 f 3N2 + 6H2O (16)
concepts described in assumptions V and VII. The
equilibrium conditions of the reactions in the different 4NH3 + 6NO f 5N2 + 6H2O (17)
pathways allow manipulation of the equations as shown
in the figure. In this way, we have accounted for the Equations (18)-(20) describe the forward rates of
complex equilibria of the three distinct pathways by reactions (15)-(17), respectively, and Table 2 sum-
applying the above assumptions. marizes the corresponding parameter values. The re-
After assumptions I-VII are applied, the overall
reaction mechanism simplifies to the one shown in -4d[NO]/dt ) A1e-E1/RTpNOapO2bpNH3c (18)
Figure 4. We note that the validity of the simplified
reaction set proposed here depends on a high liquid
residence time at each column stage, as is the case in
-2d[NO2]/dt ) A2e-E2/RTpNO2xpO2ypNH3z (19)
tray absorbers. The resulting approximation of the
liquid-phase reaction equilibrium allows the application -4d[NO]/dt ) A3e-E3/RTpNOmpNH3n (20)
of assumptions III, V, and VII. In the cases of packed
and spray absorption towers, the inaccuracy of the verse of reaction (15) accounts for the effect of water on
equilibrium approximation requires the treatment of the SCR reactions, as described by Willi et al.9 Kinetic
hydrolysis reactions (4), (8), and (13) with the decom- equation (21) mathematically describes this reaction.
position reaction (5) as kinetic reactions.
ASPEN PLUS calculates vapor-liquid equilibrium in 4d[NO]/dt ) A4e-E4/RTpH2OdpN2e (21)
the absorber, so all species are capable of interfacial
mass transport. However, by far the most important 6. Absorption Simulation Results
mass-transfer component is N2O4, which must absorb
into the aqueous phase before reacting. Also note that 6.1. Comparison to Design Specifications. Tables
excess NO can, by mass action, absorb and push reaction 3 and 4 compare the model results for key flow streams
2622 Ind. Eng. Chem. Res., Vol. 40, No. 12, 2001

Table 3. Gas- and Liquid-Outlet Streams from the Absorber

data source RFAAP data model RFAAP data model
stream fume from top fume from top acid from bottom acid from bottom
component mole flow (concentration) kg‚mol/h (ppmv) kg‚mol/h (ppmv) kg‚mol/h kg‚mol/h
NO 0.63 (2127) 0.65 (2195)
NO2 0.12 (405) 0.12 (401)
total NOx 0.75 (2532) 0.76 (2598)
HNO3 trace 0.00 0.95 0.937
O2 59.9 60.0
N2 225.6 225.6
H2O 9.9 9.2 5.9 6.4
total flow, kg‚mol/h 296.2 295.5 9.4 8.24
total flow, kg/h 8440.4 8430.5 170.5 190.9
temperature, °C 26.7 26.0 30.0 31.4
pressure, kPa 96.5a 96.5a 413.7b 413.7b
acid wt % 35.2 30.9
a Column run at ambient pressure. High-altitude operation dictates a pressure lower than 101.325 kPa. b Bottoms pressure measured
at acid pump discharge.

Table 4. Stream Results for the Kinetic Model As

Compared to RFAAP Data
data source RFAAP data kinetic model
stream gas to vent stack gas to vent stack
(SCR output) (SCR output)
component mole flow kg‚mol/h (ppmv) kg‚mol/h (ppmv)
(concentration)
NO 0.027 (91) 0.028 (96)
NO2 trace 0.002 (4)
total NOx 0.027 (91) 0.031 (100)
NH3 trace 0.037
O2 59.9 59.9
N2 225.3 226.3
H2O 10.8 10.4
total flow, kg‚mol/h 297.0 296.6
total flow, kg/h 8450.9 8445.0
temperature, °C 177 177
pressure, kPa 97 103 Figure 6. Comparison of absorption results to those of Thomas
Note: The kinetic model uses the same input NH3 as RFAAP and Vanderschuren.8
at 0.621 kg-mol/hr.
range of the absorber at RFAAP passes directly through
this maximum. The shift in this maximum with respect
to different process variables produces either a positive
or negative effect of the inlet NOx concentration on
removal efficiency.
Physically, this maximum marks the transition be-
tween absorption limited by the gas concentration to
that by the liquid concentration. At low NOx concentra-
tions, the concentrations of vapor-phase reaction prod-
ucts are low. Thus, the rate of absorption of these
species proceeds slowly. As NOx concentrations increase,
more NO2 exists as the more readily absorbed N2O4.
Eventually, the equilibrium nitric acid concentration in
the scrubbing liquid begins to limit absorption as the
backward reactions and desorption of species from the
liquid increase. The location of this maximum can, in
fact, move along the x axis. Many variables, including
Figure 5. Plot of NOx removal efficiency vs total NOx fed to the the water flow rate, temperature, and pressure, affect
column. The arrow locates the inlet NOx concentration (5200 ppmv) the location of this transition point.
to the absorber at RFAAP. A second key area for comparison is the oxidation
ratio, which is essentially the ratio of NO2 and N2O4 to
to the RFAAP design specifications. As the tables show, the total NOx. Figure 6 compares the model’s perfor-
the simulation agrees quite closely with design speci- mance to results presented by Thomas and Vanders-
fications. churen.8 We compared a single equilibrium stage in our
6.2. Comparison to Literature Data. Direct com- model to the results of a small bench-scale packed
parisons for simulation results for NOx absorption with column used by Thomas and Vanderschuren. We use
literature data remain elusive because of the number temperatures, pressures, and inlet NOx concentrations
of variables affecting the system and the difficulty of identical with those of the experimental column for our
locating literature data with similar experimental con- simulation. The gas residence times for the simulation
ditions. For example, the removal efficiency versus inlet and the experimental column were comparable with 3.0
NOx concentration (ppmv), shown in Figure 5, exhibits and 2.6 s, respectively. The model’s sensitivity to the
a clear maximum in the data. In fact, the operating NOx oxidation ratio agrees with the literature. We see
 Ind. Eng. Chem. Res., Vol. 40, No. 12, 2001 2623

Figure 9. Plot of NO2 out the top of the column vs filtered water
fed to the column for varying NO2 concentration in fume (1 m3/h,
Figure 7. Simulation results for vapor-component stage composi- 4.4 gpm).
tions for the absorber for the equilibrium model. Top ) 1. Bottom
) 18.

Figure 10. Plot of NO2 out of the top of the column vs fume feed
temperature to column (simulating a cooling jacket).

Figure 8. Simulation results for liquid-component stage composi- A second option involves temperature control: reduc-
tions for the absorber for the equilibrium model. Top ) 1. Bottom
) 18.
ing the absorber temperature improves NOx absorption
dramatically. If the operator can control the tempera-
ture of the fume feed, a feedback control loop can
that, at the 100% oxidation ratio, the NOx absorption
monitor the NOx content exiting in the gas stream and
efficiency approaches the “2/3” limit described by Cher-
cool the fume feed to the temperature required to
emisinoff and Young.1 This limit derives from the
achieve the desired NOx outlet concentration, as shown
stoichiometry of reaction (4-5). If the reaction proceeds
in Figure 10. This figure shows NO2 inlet rates to the
to completion, then, for every 3 kg‚mol of NO2 (1/2N2O4)
column of 1.5, 1.8, 2.3, and 2.7 kg‚mol/h.
absorbed, 1 kg‚mol of NO is produced, thereby ef-
In the case of RFAAP, the fume feed rate greatly
fectively yielding only 2 kg‚mol of overall NOx absorp-
exceeds the water feed rate to the column. Therefore,
tion. In actuality, equilibrium ensures that this limit is
the water feed to the column has a negligible effect on
approached but never reached.
the column temperature relative to the fume feed.
6.3. Practical Implications. Figures 7 and 8 show Adding a heat exchanger to the fume feed stream is a
the vapor- and liquid-composition profiles, respectively, less expensive retrofit than adding a cooling jacket or
for each of the 18 stages. Most of the NOx absorption cooling trays to an existing column. New designs should
(identified by the change in concentration) occurs at the explore the use of these alternatives for column cooling,
top of the column, where the pure filtered water feed especially if the water flow rate is low relative to the
creates the greatest driving force. Water flowing down gas flow rate.
through the column becomes increasingly fortified in
nitric acid until it reaches a limiting concentration that
7. SCR Simulation Results
precludes further absorption. Hence, little additional
absorption occurs below stage 10. 7.1. Optimum Ammonia-to-NOx Feed Ratio. The
In practice, spikes in the NOx content of the absorber fume outlet of the absorber, after heating and mixing
feed inevitably occur. The simplest remedy is to increase with ammonia, becomes the feed to the SCR catalyst
the freshwater feed to the tower, diluting the nitric acid vessel. Key factors in SCR performance include the NOx
in the bottom and increasing the driving force for NOx concentration, ammonia-to-NOx feed ratio, and oxygen
absorption. Theoretically, a new feedback controller feed concentration. Because the catalyst vessel temper-
could add water based on the NO2 outlet from the tower, ature is tightly controlled between 305 and 325 °C (both
as shown in Figure 9. for safety and for avoidance of undesirable reactions),
2624 Ind. Eng. Chem. Res., Vol. 40, No. 12, 2001

Figure 13. Plot of the effect of oxygen in the feed on the fractional
conversion of NO for the simulation presented in comparison to
Figure 11. Plot of the fraction of the feed that escapes the catalyst data of Willi et al.9
vessel versus the ammonia-to-NOx feed ratio.

of use as well as contamination from acids and liquid

water degrade the catalyst performance considerably.
Consequently, the catalyst activity level must be tested
periodically to predict the performance and schedule
refurbishment.
Figure 13 shows the effect of oxygen on the feed for
dry and wet gas streams in both the model and Willi et
al.9 Again, the model’s trend matches the reported
results, with the error considered a small shift down-
ward in performance. The discrepancy is largest at low
oxygen concentrations, where its effect is greatest. The
error stems from differences between the model and the
literature in terms of both NOx feed concentration and
catalyst vessel size, though again in both cases the
results are for a pristine catalyst.
The SCR at RFAAP runs at 20% oxygen in the gas
Figure 12. Plot of the effect of water in the feed on the fractional feed, safely above the point of diminishing returns,
conversion of NO compared to the results of Willi et al.9 though hardly by design. As noted previously, large
quantities of entrained air hinder the absorber perfor-
it has little effect on performance over this narrow
mance upstream. For situations with limiting oxygen
range.
levels, any air fed to the catalyst vessel should be
Figure 11 shows the effect of the ammonia feed ratio
minimized.
on the NOx reduction efficiency for the SCR model.
Optimal NOx removal occurs at an NH3-to-NOx feed
ratio of 1.2. Typically, the optimum ratio is 1.0; however, 8. Retrofit Options
because 1 mol of NO2 requires 2 mol of NH3 to react,
appreciable amounts of NO2 in the system boost the NH3 Aside from very expensive catalyst refurbishment, the
requirement so that mixtures of NO and NO2 require a SCR at RFAAP is running at near optimum conditions,
ratio greater than 1.0. The optimum ratio can only be and the strict safety limits on temperatures and the
determined by studying each system; in all cases, ammonia feed ratio preclude most retrofit design op-
though, strict feedback control should be used to mini- tions. Therefore, the following options directly address
mize both NOx and ammonia slip. only the absorber, with downstream effects noted where
7.2. Comparison to Literature Data. Figure 12 applicable.
compares the model’s response to water in the SCR feed 8.1. Column Cooling. Cooling the column improves
to results published by Willi et al.9 Although the model nearly all aspects of NOx absorption. Remarkably, all
shows a systematic deviation, it does predict the general important equilibrium reactions are exothermic. Reac-
trend and provides an effective approximation in the tion (1), the only rate-limiting reaction, actually pro-
relevant operating range. Importantly, water in the SCR ceeds faster as the temperature decreases (quite unique
feed stream appears to be an issue at RFAAP; the arrow in the study of kinetics). Therefore, cooling the process
in Figure 12 marks the current water content (3.6% by has few disadvantages. The primary problem is finding
mole). a feasible way to apply the theory. The most promising
This problem, stemming from the upstream water method appears to be cooling the fume stream with
absorber, highlights the importance of viewing the chilled water or using an acid-resistant chiller to cool
entire abatement system holistically. Cooling the ab- the high-flow-rate scrubber liquid recycle. Figure 14
sorber column would condense more water and, conse- shows the simulation results for the first option. With
quently, pass less to the SCR. RFAAP should consider a cooled gas stream, the downstream heaters must
this option, especially if Willi et al.9 are correct in that provide additional heat to maintain the optimal SCR
some of the effects of water represent irreversible reaction temperature. This heat can easily come from
damage to the catalyst. Equally important, while Figure increasing either the steam to the steam preheater or
12 presents results for a fully functional catalyst, years the natural gas to the direct-fired heater.
 Ind. Eng. Chem. Res., Vol. 40, No. 12, 2001 2625

Figure 14. Block flow diagram of the NOx abatement system utilizing fumes cooling.

Column cooling also dilutes the recoverable acid by 9. Retrofit Economics

condensing water that otherwise would leave the col-
Table 5 presents the economics for the key retrofit
umn in the gas stream, creating a second economic
options relative to the base case/current situation at
drawback. This problem is important for nitric acid
RFAAP. When economics are evaluated, however, it is
plants where acid is easily recovered but less so for important to remember that traditional payback times
applications that do not have the infrastructure for may have little relevance in pollution-control discus-
recovery. sions. Rules and regulations of the EPA and VDEQ,
Finally, column cooling also significantly benefits the along with the fines for noncompliance, change con-
downstream SCR. With more water condensing and stantly. Consequently, it may be misleading to calculate
leaving in the liquid bottoms, less water enters the payback times solely from assumptions for the required
catalyst vessel, which improves the SCR reaction (Fig- engineering design and skilled labor work plus esti-
ure 12) and reduces catalyst damage.9 mates for the cost of piping, equipment, and materials.
8.2. Acid Flash. Although the relatively low temper- Cooling the fume feed is a case in point; this retrofit
atures in the SCR (as well as the relative isolation of has no payback time because the annual costs are
the SCR at RFAAP) preclude heat integration, the stack higher than those of the base case. Yet, such a retrofit
gas offers a ready source of energy for the NOx abate- may prove invaluable in meeting EPA and VDEQ
ment system. One novel way to handle the effects of requirements.
additional water on acid recovery, for example, is with In other words, the absence of a direct payback period
a flash drum to concentrate the acid. Figure 15 shows for the retrofit does not take into account the insurance
the simulation results for this scenario. At RFAAP, the it provides against fines levied for excess NOx emissions.
minimum acid concentration offering feasible recovery Should the base-case performance prove subpar, it does
is 30 wt %; as the results show, a single-stage flash not take many $25 000/day fines to make the retrofit
drum can achieve this concentration with no other more economically feasible. One thing we know for
modifications beyond a lower filtered water require- certain is that governmental agencies are not relaxing
ment. their emission limits on NOx and other pollutants; these
The costs are piping, equipment, and transportation limits will only become tighter in the future, and
of the recovered acid to the nitric acid plant on site. The companies need to be prepared to handle the restrictions
added benefit is that controlling the filtered water flow if they want to avoid costly fines and penalties.
rate becomes less critical, making it easy to add water
to cover fluctuations in NOx input. In addition, if 10. Observations for NOx Abatement Utilizing
desired, the gas flow from the top of the absorber can Absorption and SCR
cool the water vapor leaving the flash drum as well as Using equilibrium reactions to model NOx absorption
slightly preheat the gas stream destined for the SCR. and using kinetic reactions to model the SCR agree
A multiple-stage flash would minimize the filtered water sufficiently with both design specifications and litera-
requirement. ture data for the conditions encountered at RFAAP.
2626 Ind. Eng. Chem. Res., Vol. 40, No. 12, 2001

Figure 15. Block flow diagram of the NOx abatement system implementing acid flash.

Table 5. Process Retrofit Economics11,12

base case I: cooling the fume stream II: acid flash
capital investments/costs/recoveries amount cost, $ amount cost, $ amount cost, $
Equipment/Fixed Costs
pump 1095 1095
tank
heat exchanger 16428 3286
chiller 13142
column mod.
H2O2 storage/plant
piping, installed (m) 100 4922 100 4922
engineering (h) 80 4000 80 4000
skilled labor (h) 80 2000 80 2000
total fixed costs 41587 15302
Annual Operating Costs
process water (cm3) 1.40 × 103 404 8.00 × 102 231 2.30 × 103 658
slaked lime (kg) 5.30 × 105 24861 5.30 × 105 24917 5.30 × 104 2492
cooling water (cm3) 3.70 × 106 135983 2.30 × 105 8567
refrigeration (mJ) 9.20 × 1010 666
electricity (kW‚h) 1.80 × 105 8635
steam (kg) 4.70 × 105 2681
total annual operating costs 25265 170432 14398
recovered acid (kg) 4.70 × 105 2499
annual costs minus recoveries 25265 170432 11899
payback time (years) 1.14
(i) Controlling the temperature, pressure, and filtered- to investigate ways to improve mass transfer, such as
water feed rate can greatly improve NOx absorption in replacing bubble-cap trays with sieve trays.
the scrubber absorber. (v) Diluting NOx in an inert gas slows its absorption
(ii) These dramatic improvements in NOx absorption in the absorber and its reduction in the SCR because of
(pressure, cooling, and increased water input) dilute the
concentration-dependent reactions in both units. Con-
bottom acid product.
sequently, although oxygen improves the SCR reaction
(iii) Each temperature and pressure state within the
column has a limiting NOx absorption based on a nitric downstream, entrained air in NOx gas destined for
acid concentration above which no NOx absorbs. absorption should be minimized.
(iv) Because slow mass-transfer rates represent a (vi) Oxygen is important to successful SCR operation,
major limiting step in NOx absorption, it is important and although its contribution to NOx absorption is
 Ind. Eng. Chem. Res., Vol. 40, No. 12, 2001 2627

minimal in this case, it becomes more important at pN2 ) N2 partial pressure, kPa
higher NO partial pressures. pNH3 ) NH3 partial pressure, kPa
(vii) Water has a marked deleterious effect on SCR pNO ) NO partial pressure, kPa
operation, both transient and permanent. Preceding the pNO2 ) NO2 partial pressure, kPa
SCR with absorption increases the risk of water damage pO2 ) O2 partial pressure, kPa
to the SCR catalyst. Designers should take steps to ppmv ) parts per million by volume
minimize water intrusion to the catalyst vessel. scfm ) standard cubic feet per minute, ft3/min
Beyond developing specific conclusions, however, the wt % ) HNO3 absorber liquid outlet weight percent, kg of
practical reaction mechanism presented, built using HNO3/kg of the total
x ) reaction order with respect to NO2, eq 19
readily available process simulation software, should
y ) reaction order with respect to O2, eq 19
offer a sense of empowerment. Ideally, the method for
z ) reaction order with respect to O2, eq 19
developing the mechanisms for NOx absorption and SCR
transcends these two processes alone; the logic and
information here can serve as a starting point for Literature Cited
developing usable models for other situations, again, (1) Cheremisinoff, P. N., Young, R. A., Eds. Air Pollution
always taking the particulars of each scenario into Control and Design Handbook; Marcel Dekker: New York, 1977;
account. Part 2, pp 628 and 671-678.
(2) Cho, S. M. Properly Apply Selective Catalytic Reduction for
NOx Removal. Chem. Eng. Prog. 1994, 90 (No. 1), 39.
Acknowledgment (3) Miller, D. N. Mass Transfer in Nitric Acid. AIChE J. 1987,
33, 1351.
We gratefully acknowledge the support of this work (4) Suchak, N. J.; Joshi, J. B. Simulation and Optimization of
by the Alliant Techsystems, contract operator of the NOx Absorption System in Nitric Acid Manufacture. AIChE J.
Radford Facility and Army Ammunition Plant, Radford, 1994, 40, 944.
VA. We also thank the Honeywell International Foun- (5) Matasa, C.; Tonca, E. Basic Nitrogen Compounds: Chem-
dation and Aspen Technology, Inc. (in particular, Dr. istry, Technology and Applications, 3rd ed.; Chemical Publishing
Jita Mahalec, Director, Worldwide University Pro- Co., Inc.: New York, 1973; p 44.
(6) Thomas, D.; Vanderschuren, J. Modeling of NOx Absorption
grams), for supporting the computer-aided design edu- into Nitric Acid Solutions Containing Hydrogen Peroxide. Ind.
cational program at Virginia Tech. Eng. Chem. Res. 1997, 36, 3315.
(7) Newman, B. L.; Carta, G. Mass Transfer in the Absorption
Nomenclature of Nitrogen Oxides in Alkaline Solutions. AIChE J. 1988, 34, 1190.
(8) Thomas, D.; Vanderschuren, J. The Absorption-Oxidation
a ) reaction order with respect to NO, eq 18 of NOx with Hydrogen Peroxide for the Treatment of Tail Gases.
A1 ) preexponential factor, eq 18, kg‚mol/(s‚m3) Chem. Eng. Sci. 1996, 51, 2649.
A2 ) preexponential factor, eq 19, kg‚mol/(s‚m3) (9) Willi, R.; Roduit, B.; Koeppel, R. A.; Wokaun, A.; Baiker,
A3 ) preexponential factor, eq 20, kg‚mol/(s‚m3) A. Selective Reduction of NO by NH3 over Vanadia-Based Com-
A4 ) preexponential factor, eq 21, kg‚mol/(s‚m3) mercial Catalyst: Parametric Sensitivity and Kinetic Modeling.
Chem. Eng. Sci. 1996, 51, 2897.
b ) reaction order with respect to O2, eq 18 (10) Marangozis, J. Comparison and Analysis of Intrinsic
c ) reaction order with respect to NH3, eq 18 Kinetics and Effectiveness Factors for the Catalytic Reduction of
CMM ) cubic meters per minute, m3/min NO with Ammonia in the Presence of Oxygen. Ind. Eng. Chem.
d ) reaction order with respect to H2O, eq 21 Res. 1992, 31, 987.
e ) reaction order with respect to N2, eq 21 (11) Peters, M. S.; Timmerhaus, K. D. Plant Design and
E1 ) activation energy, eq 18, kcal/g‚mol Economics for Chemical Engineers, 4th ed.; McGraw-Hill: New
E2 ) activation energy, eq 19, kcal/g‚mol York, 1991; pp 661-677.
E3 ) activation energy, eq 20, kcal/g‚mol (12) Marshall and Swift Equipment Cost Index. Chem. Eng.
E4 ) activation energy, eq 21, kcal/g‚mol 1999, May.
Ev ) vaporization efficiency
Received for review June 1, 2000
m ) reaction order with respect to NO, eq 20
Revised manuscript received March 21, 2001
n ) reaction order with respect to N2, eq 20 Accepted March 26, 2001
NO2* ) total NO2 species (NO2 + N2O4)
pH2O ) H2O partial pressure, kPa IE0005295