168
Research Collection Doctoral Thesis Simulation of NOx reduction over a Fe-Zeolite catalyst in an NH3- SCR system and calibration of the related parameters Author(s): Sharifian, Leila Publication Date: 2011 Permanent Link: https://doi.org/10.3929/ethz-a-006716204 Rights / License: In Copyright - Non-Commercial Use Permitted This page was generated automatically upon download from the ETH Zurich Research Collection . For more information please consult the Terms of use . ETH Library

Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

  • Upload
    others

  • View
    17

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Research Collection

Doctoral Thesis

Simulation of NOx reduction over a Fe-Zeolite catalyst in an NH3-SCR system and calibration of the related parameters

Author(s): Sharifian, Leila

Publication Date: 2011

Permanent Link: https://doi.org/10.3929/ethz-a-006716204

Rights / License: In Copyright - Non-Commercial Use Permitted

This page was generated automatically upon download from the ETH Zurich Research Collection. For moreinformation please consult the Terms of use.

ETH Library

Page 2: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Doctoral Dissertation

DISS. ETH NO. 19736

Simulation of NOx reduction over a Fe-Zeolite catalyst

in an NH3-SCR system

and calibration of the related parameters

Leila Sharifian

Aerothermochemistry and Combustion Systems Laboratory (LAV)

Department of Mechanical and Process Engineering (D-MAVT)

Swiss Federal Institute of Technology Zurich (ETHZ)

Page 3: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären
Page 4: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

DISS. ETH NO. 19736

Simulation of NOx reduction over a Fe-Zeolite catalyst

in an NH3-SCR system

and calibration of the related parameters

A dissertation submitted to

ETH ZURICH

for the degree of

Doctor of Sciences

presented by

LEILA SHARIFIAN

Master of Science in Mechanical Engineering, Sharif University of Technology

born September 16th, 1980

citizen of Iran

accepted on the recommendation of

Prof. Dr. Konstantinos Boulouchos, ETH Zürich, examiner

Dr. Oliver Kröcher, Paul Scherrer Institute, co-examiner

Dr. Yuri Martin Wright, ETH Zürich, co-examiner

2011

Page 5: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären
Page 6: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

To Ebrahim and my parents

Page 7: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären
Page 8: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Preface

The present work was carried out at the Laboratory of Aerothermochemistry and Com-

bustion Systems (LAV) at the Swiss Federal Institute of Technology (ETH) Zurich,

Switzerland as a part of the CCEM-NEADS project funded by the Swiss Competence

Centre for Energy and Mobility and the Swiss Federal Office of Energy (BFE).

First my utmost gratitude to Prof. Dr. K. Boulouchos, head of LAV, whose supports and

encouragements from the preliminary to the concluding level enabled me to manage the

biggest vicissitude ever in my life. My very special and hearty thanks to my supervisor

Dr. Y.M. Wright whose sincerity and supervision helped me to hurdle all the obstacles

in the completion of this research work.

I’d like to specially appreciate Dr. O. Kröcher for his valuable scientific support and the

successful cooperation between PSI and us which makes this research possible. My re-

spectful thanks go to Mr. M. Elsener for providing the experimental data of the system

and Dr. T. Schildhauer for the helpful discussions. I’d like to sincerely thank Dr. J.

Mantzaras and Dr. Ch. Frouzakis for the worthful guidance and comments whenever I

needed.

This thesis would not have been possible without the excellent working environment in

LAV and I warmly gratitude all my colleagues whom I shared the friendly atmosphere

with. My special thanks go to Ms. S. Blatter and Ms. F. Meyer for pursuing the compli-

cated administration procedure and their helps particularly in my first days.

Last but not the least I’d like to thank my parents and family for their supports in whole

my life and their morale when I was far from home. I share the credit of my work with

my lovely husband the one who donates me fortitude in hard days. I deeply gratitude

Ebrahim for the safe, peace and serenity brings to me. And the one above all of us,

thanks the omnipresent God for his gifts.

Leila Sharifian

Zurich, June 2011

Page 9: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären
Page 10: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Summary

NOx reduction in an ammonia SCR converter is simulated by the own developed code

and a calibration procedure is proposed to determine the system parameters. A 1D+1D

model is carried out for a single representative channel to parametrically study the char-

acteristics of the system under typical operating conditions. An appropriate model is

developed interpreting the chemical behavior of the system and the parameters specify-

ing the characteristics of the SCR system are introduced. An efficient calibration proce-

dure is proposed in an exploiting structure of the experiments with an Fe-Zeolite

washcoat monolith for different feed concentrations, temperatures and flow rates. Phys-

ical and chemical properties are determined as well as kinetics and rate parameters and

the model is verified by experimental data at different operating conditions.

Three different mechanisms for the surface kinetics to model NOx reduction are as-

sessed and the results are compared in the cases of steady DeNOx performance and tran-

sient response of the system. Ammonia inhibition is considered in the model since it has

a major effect specifically under transient operating conditions. Effects of the operating

temperature, the gaseous flow rate and the species concentrations such as ammonia dos-

age and the ratio of NO and NO2 are investigated. It is shown that recently proposed

dual site chemical kinetics for Fe-Zeolite catalysts are especially suited to predict the

transient operation of the system during ammonia feeding and the inhibition effect after

shutting off ammonia.

NOx reduction is simulated as well for highly transient conditions representative of the

automotive Diesel engine operation. A set of dynamic experiments in totally 90 cases of

space velocities, temperatures and concentrations is used to investigate and validate the

system. It is shown that the model has a strong capability to predict the performance of

the SCR system for steady and transient conditions as well as highly transient operation.

Page 11: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

IV Summary

Accurate numerical results for various cases are a strong indication for the validity of

the model under the typical operating conditions of an SCR system. This study helps to

find the optimum performance of the system under different conditions including a wide

range of working temperatures with different flow rates and concentrations.

Page 12: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Zusammenfassung

In der vorliegenden Arbeit wird die NOx-Reduktion in einem Ammoniak-SCR-

Konverter mit einem eigens entwickelten Code simuliert, und weiterhin eine

Kalibrierung vorgeschlagen, über welche die System-Parameter bestimmt werden

können. Ein 1D+1D-Modell wird für einen representativen Kanal verwendet und die

Merkmale des Systems werden parametrisch unter typischen Einsatzbedingungen

geprüft. Dabei wird das chemische Verhalten des Systems interpretiert und die

Parameter zur Spezifikation der Eigenschaften des SCR-Systems vorgestellt. Eine

effiziente Kalibrierungsmethode wird über eine Reihe von Experimenten mit einem mit

Fe-Zeolith beschichteten Washcoat-Monolith vorgeschlagen, bei der verschiedene

Konzentrationen, Temperaturen und Massenflüsse variiert werden. Somit werden

sowohl die physikalischen und chemischen Eigenschaften als auch die Kinetiken und

die Parameter der Reaktionsrate bestimmt. Die Ergebnisse des Modells sind anhand von

mehreren Betriebspunkten validiert.

Die NOx-Reduktion wird anhand von drei verschiedene Mechanismen Kinetik

modelliert und die Ergebnisse unter stationären und transienten Bedingungen

verglichen. Dabei ist im Model eine Inhibiting durch Ammoniak berücksichtigt, welche

speziell bei transienten Bedingungen einen großen Einfluss aufweist. Weiterhin werden

die Auswirkungen der Betriebstemperatur, der Strömungsgeschwindigkeit, der Einfluss

der Ammoniakdosierung und das Verhältnis von NO und NO2 untersucht. Dabei wird

gezeigt, dass mit einer kürzlich veröffentlichten dual site chemischen Kinetik für Fe-

Zeolith-Katalysatoren aus der Literatur das transienten Verhalten Systems bei Ein-und

Abschalten der Ammoniumzuführung besonders gut vorhersagt.

Die NOx-Reduktion wird auch für hoch transiente Bedingungen simuliert, wie sie bei

einem Dieselmotorischen Prozess verkommen. Dabei wird eine Reihe von 90

dynamischen Versuchen mit Variationen von Raum-Geschwindigkeiten, Temperaturen

und Konzentrationen verwendet, um das System zu untersuchen und zu validieren.

Page 13: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

VI Zusammenfassung

Es konnte gezeigt werden, dass das Modell die Eigenschaften einen SCR-Katalysators

sowohl für stationäre als auch bei transienten Bedingungen sehr gut vorhersagt.

Die gute Übereinstimmung für verschiedene Fälle ist ein starkes Indiz für die Validität

des Modells im Rahmen typischer Betriebsbedingungen eines SCR-Systems. Somit hilft

diese Studie die optimale Leistung des System bei verschiedenen Bedingungen,

einschließlich einer grossen Bandbreite von Betriebstemperaturen mit verschiedenen

Strömungsgeschwindigkeiten und -Konzentrationen zu finden.

Page 14: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Contents

Chapter 1. Introduction ................................................................................................ 1

1.1 NOx emissions and Diesel engines .................................................................... 1

1.2 Selective Catalytic Reduction (SCR) ................................................................. 4

1.2.1 Upstream process ........................................................................................ 4

1.2.2 Heterogeneous reactions ............................................................................. 5

1.2.3 Catalytic converter ...................................................................................... 6

1.3 Objective of this study ....................................................................................... 7

Chapter 2. State of the art ............................................................................................. 9

2.1 Simulation of an Ammonia-SCR system ........................................................... 9

2.1.1 Mass and heat transfer simulation in a catalytic converter ....................... 10

2.1.2 Kinetics of selective catalytic reduction of NOx by ammonia.................. 16

2.2 Conclusion ....................................................................................................... 22

Chapter 3. Methodology ............................................................................................. 29

3.1 Modeling of the monolith converter ................................................................ 29

3.1.1 The converter channel .............................................................................. 29

3.1.2 Catalytic washcoat layer ........................................................................... 30

3.1.3 Boundary condition at the catalytic wall .................................................. 32

3.2 Kinetics of catalytic heterogeneous reactions .................................................. 34

3.2.1 Ammonia adsorption/desorption/spill-over .............................................. 35

3.2.2 Standard and fast SCR .............................................................................. 37

3.2.3 NH3 oxidation ........................................................................................... 38

3.2.4 NO oxidation ............................................................................................ 39

3.2.5 NO2 and NH3 reactions ............................................................................. 39

3.3 Heat of reactions .............................................................................................. 40

3.4 Transient 1D+1D model including heterogeneous reactions ........................... 41

3.5 Numerical method ............................................................................................ 41

Chapter 4. Experiment ............................................................................................... 47

Chapter 5. Calibration procedure and results ............................................................. 53

Page 15: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

VIII Contents

5.1 Grid independency ........................................................................................... 53

5.2 Wall diffusion .................................................................................................. 55

5.3 Calibration procedure ...................................................................................... 57

5.3.1 Ammonia adsorption/desorption .............................................................. 59

5.3.2 NO oxidation ............................................................................................ 64

5.3.3 Model I and II: Standard and fast SCR reactions ..................................... 67

5.3.4 Model III: Standard and Fast SCR and ammonia spill-over .................... 70

5.3.5 NH3 oxidation and reactions between NO2 and NH3 ............................... 73

5.4 Transient operation and steady state ................................................................ 74

5.4.1 Model II .................................................................................................... 74

5.4.2 Model III ................................................................................................... 77

5.5 Highly transient performance .......................................................................... 86

5.6 Non-isothermal operation ................................................................................ 95

5.7 Conclusions .................................................................................................... 100

Chapter 6. Discussion and outlook ........................................................................... 102

6.1 Discussion ...................................................................................................... 102

6.2 Outlook .......................................................................................................... 104

Nomenclature................................................................................................................ 107

Referred publication ..................................................................................................... 111

References ................................................................................................................ 113

Appendix ................................................................................................................ 121

Page 16: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 1. Introduction

1.1 NOx emissions and Diesel engines

The two oxides of Nitrogen that are of primary concern to air pollution are NO and

NO2. NO is a colorless gas that is a precursor to NO2 and is an active compound in pho-

tochemical reactions that produce smog; while NO2 is a reddish brown gas that gives

color to smog [1, 2]. NO2 is a criteria pollutant with a National Ambient Air Quality

Standard (NAAQS) of 100 μg/m3 or 0.053 ppm, annual average [3]. It is also a precur-

sor to nitric acid, HNO3, in the atmosphere and is a major contributor to acid rain. Final-

ly, NOx and volatile organic compounds (VOC) react photochemically in a complex se-

ries of reactions to produce smog, which includes ozone, NO2, peroxyacetyl nitrate

(PAN), peroxybenzoyl nitrate (PBN) and other trace oxidizing agents [4].

Table 1.1. European [5] and the US [6] standards for heavy duty Diesel engines NOx emission

Legislation Implementation Date NOx (g/kWh)

Euro I Jan. 1992 8.00

Euro II (a) Oct. 1996 7.00

Euro II (b) Oct. 1998 7.00

Euro III Oct. 2000 5.00

Euro IV Oct. 2005 3.50

Euro V Oct. 2008 2.00

Euro VI Jan. 2013 0.40

EPA 04 Jan. 2004 3.33

EPA 07 Jan. 2007 1.80

EPA 10 Jan. 2010 0.27

Page 17: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

2 Chapter 1. Introduction

By far the largest source of NOx is combustion, although there are other industrial

sources such as nitric acid manufacturing. A very large part of the contribution of NOx

is generated by motor vehicles and other transportation, including ships, airplanes, and

trains. NOx is produced during combustion by three mechanisms, “thermal”, “prompt”,

and “fuel NOx”. In an internal combustion engine, combustion temperature is high

enough to drive endothermic reactions between Nitrogen and Oxygen yielding oxides of

Nitrogen such as NO and NO2. [1]

To prevent an increase in total NOx emissions, different legislations have been intro-

duced and standards are being tightened to control NOx produced by automobiles. Table

1.1 reported the standards for NOx emission of heavy duty Diesel engines in European

Union [5, 7] and the United States [6, 8].

Figure 1.1. NOx and PM trade-off in European emission standard.

Two basic strategies exist to attain the newest standards of NOx emission. The first is

optimizing the combustion with respect to a low emission of NOx such as reduction of

the flame temperature, which leads to a high emission of unburned material (soot, CO

and hydrocarbons); consequently using a particulate filter (DPF) in the after-treatment is

necessary to remove particulate matter (PM). Figure 1.1 presents the European stand-

Page 18: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 1. Introduction 3

ards for PM and NOx emissions and illustrates the trade-off between particulate and

NOx production in combustion [5, 9].

The second strategy is optimizing the combustion with respect to a low emission of un-

burned materials, which leads to a high emission of NOx and so a DeNOx process in the

after-treatment should be used to reduce NOx in the exhaust. This method results in a

better fuel economy so various processes have been studied in recent years for selective-

ly reducing NOx in the presence of Oxygen in lean exhaust gases [10].

Figure 1.2. Improvement of emission reduction in European standard

by using different technologies.

Figure 1.2 presents different techniques used for emission reduction as a combination of

NOx and PM reductions in European standard for heavy duty Diesel engines [5, 11].

Further improvements with respect to emission reduction to satisfy increasingly strin-

gent legislations can be obtained by applying combinations of the aforementioned tech-

niques, reviews hereto can be found e.g. in [12-15].

Current strategies attempting to minimize fuel consumption and CO2 emission prompt

the use of lean burn engine. Under condition of stoichiometric combustion, Three Way

Catalyst (TWC) is efficient and reliable and widely used to treat the automobile exhaust

of gasoline engine. However, TWC suffers from severe loss of activity for NOx reduc-

Page 19: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

4 Chapter 1. Introduction

tion in the presence of excess Oxygen, which is the prevalent condition for lean burn

engines. Thus, selective catalytic reduction (SCR) of NOx becomes a potential method

to remove NOx from exhaust and receives much attention. The reduction of NOx in lean

Diesel exhaust requires ammonia as a reducing agent which reacts selectively with NOx

in the presence of Oxygen [16]. As illustrated in Figure 1.2 the exhaust of combustion

with optimum fuel consumption can be clean enough for Euro VI standard by using

SCR DeNOx strategy and Diesel particulate filter (DPF) [17]. The lean NOx trap (LNT)

is another DeNOx technology using Diesel fuel as the reducing agent. LNTs are the pre-

ferred approach for smaller lean-burn ( Diesel, direct injection gasoline) passenger cars

[14].

1.2 Selective Catalytic Reduction (SCR)

The Selective Catalytic Reduction (SCR) process was introduced in the seventies in Ja-

pan for reducing the NOx emissions from the lean exhaust of stationary plants. Since

then, SCR has gained wide acceptance as being the most effective technology for deep

NOx removal from lean exhaust of stationary plants. SCR, using urea as reducing agent,

is presently the leading NOx reduction approach for Diesel engines. DeNOx efficiencies

can be 90% or higher with proper control and design [14]. Urea is a solid storage com-

pound of ammonia, which is actually involved in the reduction of NOx and converts

them into Nitrogen and water vapor.

1.2.1 Upstream process

An aqueous solution of urea (32.5 wt-% urea) is usually atomized into the hot exhaust

upstream of the SCR catalyst as a source of ammonia because of toxicological and safe-

ty reasons in the automotive applications. The upstream process of ammonia production

from urea solution is not the focus of this investigation and is reported according to the

literatures [18-21]. The thermo-hydrolysis of urea into ammonia and carbon dioxide

precedes the SCR reaction [10].

Page 20: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 1. Introduction 5

The first step is the evaporation of water from the urea droplets after spraying into the

exhaust, which leads to solid or molten urea,

NH2-CO-NH2 (aqueous) → NH2-CO-NH2 (molten) + x H2O (gas) (1.1)

Molten urea then is heated up and decomposed thermally,

NH2-CO-NH2 (molten) → NH3 (gas) + HNCO (gas) (1.2)

Equimolar amounts of ammonia and isocyanic acid are thus formed. Isocyanic acid is

very stable in the gas phase, but hydrolyzes easily on many solid oxides reacting with

water vapor originating from the combustion process,

HNCO (gas) + H2O (gas) → NH3 (gas) + CO2 (gas) (1.3)

The thermo-hydrolysis of urea is globally an endothermic process and the required heat

is provided by the hot exhaust gases. An optimized dosage strategy is required in order

to avoid undesired slip of ammonia and iso-cyanic acid. The main disadvantage of the

urea-SCR process is the need for a special reducing agent, which must be carried on

board of the vehicle. Urea consumption for a heavy-duty Diesel engines is reported to

amount in between 3-5% of fuel consumption [22, 23]. The mixture of ammonia and the

exhaust gases including NOx enter into the SCR converter.

1.2.2 Heterogeneous reactions

A typical SCR converter has a large number of parallel channels in a honeycomb mono-

lith structure with catalytically washcoated walls. NOx and ammonia diffuse from the

gas stream inside the channel into the pores of the washcoat where ammonia is strongly

absorbed on the catalytic surface. Absorbed ammonia reacts with the gas species inside

the pores and the products diffuse back through the pores and enter into the channel.

Page 21: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

6 Chapter 1. Introduction

The Nitrogen oxides of Diesel exhaust gas are primarily composed of NO, which con-

verts into Nitrogen and water vapor during the “standard” SCR reaction by using am-

monia in the presence of Oxygen, as follows.

4 NH3 + 4 NO + O2 → 4 N2 + 6 H2O (1.4)

DeNOx efficiency at low temperatures of SCR converters for automotive exhaust

aftertreatment is significantly enhanced by converting part of the nitric oxide to NO2,

e.g. by means of a preoxidation catalyst located upstream [24]. The reaction between

NO and NO2 and ammonia, as eq. (1.5), called “Fast” SCR reaction is faster by about

one order of magnitude than the “standard” SCR in the low temperature region (lower

than 300 ºC) over Vanadia-based catalysts [25].

4 NH3 + 2 NO + 2NO2 → 4 N2 + 6 H2O (1.5)

At high temperatures (above 300 ºC), two more reactions must be considered for ther-

mal decomposition of ammonium nitrate to N2O and the NO2-SCR reaction, in which

ammonia and NO2 react directly to produce N2 [24].

1.2.3 Catalytic converter

The different structures for converter are honeycomb and multilayer. Honeycomb cata-

lysts are in two shapes; monoliths and plates [26]. Monolith substrates are widely used

in automotive and stationary emission control reactors for the selective catalytic reduc-

tion (SCR) of Nitrogen oxides. The advantages of these catalysts are small pressure

drop and an increase in the SCR activity obtained by improving the geometric surface

area [26]. In automotive applications, ceramic monoliths are made from synthetic cordi-

erite, 2MgO.2Al2O3.5SiO2, a material having critical characteristics such as thermal

shock resistance due to a low thermal expansion coefficient; porosity and pore size dis-

tribution suitable for ease of washcoat application and good washcoat adherence; suffi-

cient strength for survival in an automotive exhaust environment; and compatibility

with washcoat and catalysts [27].

Page 22: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 1. Introduction 7

An enhanced catalytic activity is essential to increase the overall DeNOx efficiency. In

this context Vanadium-based catalysts have first been studied because of their extensive

industrial use in DeNOx processes for stationary applications. In the automotive applica-

tion, the need of high activity at low temperatures as well as activity and stability up to

higher temperatures shifts the research interest towards other catalytic systems. Conse-

quently many efforts are focused on metal-promoted Zeolites such as copper-exchanged

Zeolite and iron-exchanged Zeolite [28-31] as reviewed later in section 2.1.2.

1.3 Objective of this study

The aim of this work is simulation of the NOx reduction over a Fe-Zeolite catalyst em-

bracing broad operating conditions of SCR after treatment systems for Diesel engines,

such as temperature, space velocity, and exhaust gases mixture for highly transient con-

ditions representative of the automotive Diesel engine operation.

The monolith converter is simulated by a transient model for one single channel. Intra-

porous diffusion within the washcoated layer is simulated in the direction of the wash-

coat thickness, since the diffusion resistance limits ammonia adsorption and conse-

quently influences NOx reduction. The influence of the choice of the kinetic model and

of the operating conditions such as temperature, species concentrations and flow veloci-

ty are investigated. The chemical surface kinetics employed considers a finite rate for

the ammonia spill-over reaction by applying one of the most recent kinetics in the litera-

ture, since the strong influence of NH3 spill-over on ammonia inhibition has been identi-

fied to play a determinant role in simulation of the system under fast transient condition.

To simulate the SCR system, the parameters specifying the system characteristics are

introduced and the calibration procedure is proposed in an exploiting structure of the

experimental dataset. The simulation results are validated by experimental data for a

wide range of operating conditions in transient and steady state examining the perfor-

mance of the SCR system for different cases.

Page 23: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären
Page 24: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 2. State of the art

Development of the ammonia SCR system is a challenging task which is particularly

aggravated for mobile Diesel engines operating under highly transient conditions and

consequently producing exhaust gases with strongly fluctuating temperature, mass flow

rate, and pollutant concentration. The principle issues of each individual SCR applica-

tion are the system size and configuration as well as adequate urea injection as a source

of ammonia and control strategies [32]. The complex interaction of these issues de-

mands increasing efforts in view of design and optimization. Computer modeling helps

to make basic design decisions and therefore to shorten the overall development pro-

cess. Since simulation itself is expensive and time consuming using the right model and

computer tool for each process of interest is important for sufficient catalyst develop-

ment.

2.1 Simulation of an Ammonia-SCR system

Simulation of an ammonia-SCR system can be divided into two main parts: the catalytic

converter and its upstream. The upstream simulation includes modeling of urea spray,

evaporation, urea decomposition, ammonia formation and mixing process. This part is

not in the scope of this study and are investigated by several researches [18-21].

Simulation of the catalytic converter, which is the scope of this work, includes modeling

of mass and heat transfer and kinetics of surface reactions. Modeling of mass and heat

transfer in a catalytic converter has been developed for several application such as CO

oxidation converter, oxidation catalyst of NO, SCR and NSCR (Non-SCR). Simulation

of ammonia SCR has been specifically improved by the modeling of the surface kinet-

ics. The uncertainties of the upstream are not assessed here since a mixture of pure am-

monia and exhaust gases is used for inlet flow.

Page 25: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

10 Chapter 2. State of the art

2.1.1 Mass and heat transfer simulation in a catalytic converter

Modeling of heat and mass transfer within a monolith structure is firstly reported in

1975 [33], where a model including nondimensional differential equations describing

heat and mass transfer in a monolithic honeycomb catalyst, was developed to predict

steady states and two numerical techniques were proposed to solve the system of cou-

pled, nonlinear ordinary differential equations with split boundary conditions. A shoot-

ing procedure for integration was the first method and could be used only for problems

with low values of Peclet number which meant that conduction of heat within the wall

was more powerful than the heat convection in the channel. For high values of Peclet

number a finite difference integration approach along with the iterative solver of New-

ton-Raphson algorithm was suggested.

The presented model in [33] comprised convective heat and mass transfer in the holes of

the structure, longitudinal thermal conductivity of the honeycomb support and inter-

phase gas-solid heat and mass transfer. The equations were considered steady state in

one dimension along the length of the channel. Uniform temperatures were assumed

around the channel cross section for the gas and solid and thermodynamics properties of

the solid were approximated by the average value. The mass balance equations counted

the changes of each component along the channel and the amount of mass transfer be-

tween gas and solid phases, which was proportional to the rate of reactions in the cata-

lytic layer. Heat balance in the gas phase included temperature variations along the

channel and convective heat transfer between gas and solid wall. Heat balance in the

wall (solid phase) contained conductive heat transfer in the solid along the channels and

heat release due to the reaction coupled with convection heat transfer between gas and

solid. To estimate the values of gas-solid heat and mass transfer coefficients (Nusselt,

Nu, and Sherwood, Sh, numbers, respectively) Hawthorn’s correlation from [34] was

adopted.

It was shown that longitudinal thermal conductivity is higher in the monolithic struc-

tures than packed catalytic reactors and two stable steady states were observed for a par-

Page 26: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 2. State of the art 11

ticular monolithic structure in certain regions of operation [33]. Consequently the hon-

eycomb modules are shorter than packed beds resulted in almost zero pressure drops,

since the values of Damkohler number (Da) for honeycomb supports are lower due to

the shorter time scale of chemical reactions than convective mass transfer. Following in

a parallel study, heat and mass gas to solid transfer coefficients were experimentally es-

tablished [35] and the mass transfer results were correlated using Reynolds (Re) and

Schmidt (Sc) number and the length to diameter ratio.

Particularly for selective catalytic reduction of NOx by ammonia, two and one dimen-

sional mathematical models of monolith reactors were compared in [36]. Three different

geometries (circular, square and triangular) were considered for the channels as well as

linear and Rideal kinetics for the surface reactions discussed later in section 2.1.2. A

two dimensional analysis was carried out for steady state simulation of a single mono-

lith channel which was sufficient to represent the behavior of the whole reactor, since

conditions within each channel were identical [36]. The reactor was considered isother-

mal because of the high dilution of the reactants and the velocity profile was fully de-

veloped laminar flow in the channels. A radial diffusion equation was considered for

NO while axial mass diffusion was negligible compared to the convective contributions.

Axial molecular diffusion effects were important only at low values of the Peclet num-

ber. The reaction was considered as the boundary condition at the catalytic wall and the

balance of NO was solved numerically for different values of Damkohler Number rep-

resenting the chemical process of catalytic reaction at the wall relative to the physical

process of NO diffusional transport. This study concluded the 1D lumped parameter

model (explained later in the chapter of the methodology, section 3.1) as a promising

tool for the analysis and design of SCR reactors. Comparing different geometries, for

same Damkohler Number, the circular duct always yielded the highest Sherwood num-

ber as well as the highest conversion, and the triangular duct yielded the smallest with

the square duct in between. The reactant concentrations at the corners are lower relative

to the other wall locations away from the corners, thus the peripheral average wall con-

centration is lower for ducts with acute corners. [36]

Page 27: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

12 Chapter 2. State of the art

The lumped parameter model for the SCR reactor presented in [36] was improved by

considering the diffusion and the reactions inside the pores of the catalytic monolith

walls in [37] and the role of the intra-phase transport processes was investigated. The

proposed model including combined chemical and diffusional control successfully pre-

dicted the overall performances of the reactor with the isothermal laminar flow and the

catalytic walls. The calculations assessed the improvements of NOx removal efficiency

by modifications of the catalyst porosity and the acceptable limits of inhomogeneities in

the reactor feed stream. The importance of considering intra-porous diffusion is dis-

cussed by [38] and explained later in section 5.2.

The one-dimensional model of the SCR monolith reactor was extended to include the

treatment of the undesired SO2 oxidation reaction [39]. A redox kinetic model was used

for rate expression of SO2. The transport of NO and NH3 from the gas phase to the sur-

face of the catalyst walls and the diffusion of NO and NH3 inside the pore structure of

the catalyst were described according to [37]. SO2 was oxidized in a chemical regime

and its oxidation occurred in the whole catalytic volume since the rate of SO2 reaction

under SCR conditions was very slow. There was no inter-phase and intra-phase gradient

of SO2 and SO3 and their concentrations were constant over each cross section of the

monolith. The wall thickness of SCR monolith catalysts was partitioned between two

zones [39]. The first was a superficial layer governed by intra-porous diffusional re-

sistance where the NO and NH3 concentration gradients located and the concentration of

the DeNOx limiting reactant fell to zero. The second was an inner region where the stoi-

chiometric constraints dominated, resulting in a flat concentration profile of the excess

reactant. The second zone was the predominant one in the typical commercial SCR cata-

lysts with wall thicknesses of the order of one mm. The complete model successfully

reproduced the interaction between DeNOx and SO2 oxidation reactions. It was shown

that the catalysts with very thin walls and small channel hydraulic diameters minimized

SO2 conversion with respect to NOx reduction.

Activity of an SCR catalyst with respect to the poison accumulation was studied by a

2D model [40]. The simulations showed the possibility of improving the poison re-

Page 28: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 2. State of the art 13

sistance of SCR catalyst monoliths by reconfiguration of their pore structure and chan-

nel diameter.

The adequacy of lumped models of the catalytic combustors was analyzed in [41] by

comparison with the more detailed numerically expensive distributed ones. The models

were developed for circular and square cross section channels in steady state and fully

developed laminar flow. Mass and entropy balance equations were written in three di-

mensions cylindrical and Cartesian coordinates. The intra-porous diffusion was ac-

counted by defining the effectiveness factor while the axial diffusion was negligible.

The equations of the lumped model (one dimension) used were identical to the ones

from [37]. The local Nu and Sh numbers were calculated and compared with numbers

from the lumped model to evaluate the one dimensional analysis. It was seen, that for

long enough circular channels, the exit temperatures of the gas predicted by one dimen-

sional model were in good agreement with those provided by distributed models. It was

further concluded that in the combustor monoliths with significant extended gas phase

reactions, the lumped models might result in misleading predictions.

Adsorption and desorption of ammonia for the selective catalytic reduction of NOx were

simulated by 1D+1D model in [42] and the kinetics were studied experimentally and

theoretically by transient response techniques. The mathematical model consisted of the

material balance of adsorbed NH3 by reaction with NO, the material balances of gaseous

NO and NH3 in the porous catalyst matrix, the continuity at the gas-solid interface, and

the material balances of gaseous NO und NH3 in the monolith channel. The material

balance in the channel was written in one dimension along the channel and the equa-

tions governing the porous catalyst matrix and the gas-solid interface were in one di-

mension normal to the channels. As a result, the proposed transient system of equations

was two dimensional. The dynamics of the SCR DeNOx reaction and NH3 adsorp-

tion/desorption were investigated which is discussed bellow in section 2.1.2.

A SCR catalytic converter was simulated in [43] by a steady state lumped parameters

model for the 1D channel and the 1D diffusion within the wall. The kinetic parameters

Page 29: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

14 Chapter 2. State of the art

were determined experimentally for the SCR catalysts and the effective NO diffusion

coefficient was measured. Comparing the measured and calculated performance, the

presented model was a valuable tool to predict the influence of the parameters such as

effective diffusion coefficient, intrinsic kinetics, monolith structure, or GHSV which

helped to semi-quantitatively study the performance of the new catalysts.

In a more recent investigation [44], 1D, 2D and 3D models for the channel were com-

pared and the importance of considering diffusion within washcoat layer was reported.

A 0D model was also proposed including diffusional limitation. The authors introduced

a ratio between catalytic active surface area and geometrical surface area to use in the

transfer equations at the wall.

An equilateral triangular channel of a monolith catalytic combustor was investigated by

a 3D steady state model for heat and mass transfer and the adequacy of 1D approxima-

tion was studied [45]. Axial diffusion in the gas phase and heat transfer in the solid

phase by conduction and radiation were assumed negligible and boundary conditions at

the wall were considered with respect to the different Da for the reactions. The worst

mass and heat transfer properties were observed in triangular channels because of their

acute corners. It was shown concluded that predicted temperature profiles by simpler

one dimensional model might be significantly different.

SCR of NOx by NH3 on Vanadia honeycomb catalysts was modeled in 3D by [46] con-

sidering modeling aspects such as correct geometry representation (numerical domain as

1/8 cross section of square ducts), hydrodynamic entrance effects, adsorption phenome-

na and side reactions like direct ammonia oxidation, NH3/NO ratio in the feed, influence

of the reactant (NO, NH3, O2, H2O) concentrations, inter-phase diffusion and mass

transfer processes for a large range of operating conditions. It was concluded the 3D

model as an effective tool to design the most suitable NH3/NO feed ratio to fulfilling the

NO and ammonia emission limits.

Three alternative formulations (Navier-Stokes, boundary layer and plug flow) for simu-

lating the steady state flow and chemistry in a honeycomb channel were evaluated in

Page 30: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 2. State of the art 15

[47] for the typical conditions of catalytic combustions including homogenous reac-

tions. In the most comprehensive models, the complete Navier Stokes equations were

solved considering both axial and radial mass, momentum, and energy transport. In

models based on the boundary layer equations, axial diffusive transport was neglected

in comparison to the radial diffusion and the convective transport, but detailed transport

to and from the channel walls was retained. The plug flow model considered no diffu-

sion effect which was assumed to be small compared to axial convective transport. In

the radial direction, however, the diffusive transport (mixing) was assumed to be so

dominant that there were no radial variations in the species composition. Although the

Navier Stokes models had very few assumptions and were thus valid in the most general

setting, it was shown that they were computationally expensive. The boundary layer

models resulted accurately in moderate to high Reynolds numbers with much lower

computational cost. It was shown that, the range of the validity of the plug flow models

was limited while they were computationally inexpensive.

A steady state and isothermal mathematical model was derived with the kinetics based

upon the Eley-Rideal mechanism for the V2O5-WO3/TiO2 catalyst [48]. NO conversion

in the reactor as well as NH3 slip from it were predicted by considering the diffusion

effect on the performance of the catalytic honeycomb reactor, with respect to the cata-

lytic wall thickness, the cell size of the honeycomb, and the reaction conditions. The

mixing phenomenon of NH3 and the flue gas flow pattern were simulated within the

honeycomb reactor. The performance of a commercial scale SCR reactor is simulated

by applying 1D+1D code to the entire channels with different feed concentrations due to

ammonia distribution in the upstream. The study also identified that although the cata-

lyst layer of the reactor is rather thin, the diffusion resistance in the honeycomb reactor

is a critical key to design of the commercial scale SCR reactor.

A transient two dimensional numerical model describing the ammonia based SCR pro-

cess on Vanadia-titania catalysts was presented by [38]. The kinetics was coupled with a

fully transient two-phase 1D+1D monolith channel model. Identical conditions within

each channel of the honeycomb catalyst were assumed with negligible axial dispersion

Page 31: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

16 Chapter 2. State of the art

and pressure drop. Unsteady mass and enthalpy balances for a single monolith channel

were provided for gas and solid phase. Gas-solid mass and heat transfer resistances were

evaluated based on lumped parameter model. The strong intra-phase diffusional limita-

tions were accounted for by the equations for diffusion-reaction of the reactants in the

intra-porous field (the catalytic monolith walls). Buildup of reactants in the gas phase

within the catalyst pores was neglected. The validation with steady state and transient

experimental data revealed an excellent prediction quality. The necessity to account for

diffusional limitations in the washcoat was clearly outlined in this investigation. It was

concluded that, in general, the influence of the intraporous diffusion on the NOx conver-

sion and the NH3 slip decreased with increasing reaction temperature and decreasing gas

space velocity. Therefore the intraporous diffusion was included in the SCR reactor

model in order to describe well the effective reaction rates under all conditions varying

from chemical control to external mass transfer control. Consequently the simulation

was able for undersized up to oversized SCR catalysts within a large operating window.

2.1.2 Kinetics of selective catalytic reduction of NOx by ammonia

The modeling of ammonia SCR commenced about 20 years ago, a variety of different

mechanisms to explain the rate of reactions has been proposed in the literature. In 1990

the adsorption of ammonia and Nitrogen oxides was experimentally studied on pow-

dered Vanadia catalyst and detailed reaction mechanisms of the SCR reaction were pro-

posed [49]. It was shown that Oxygen was essential for the reoxidation of the Vanadium

sites of the catalyst since it constituted the rate limiting step in the standard SCR reac-

tion at temperatures below 300 ºC. Therefore the rate of NOx conversion of the standard

reaction, eq. (1.4), was influenced by the presence of Oxygen at low temperature.

Two years later 1D/2D simulations of circular, square and triangular channels were

compared for an SCR reactor while the reaction rates of NO reduction were modeled by

Rideal kinetics [36, 37].

The first study on the behavior of Vanadia-based SCR catalysts under transient condi-

tions was presented in [42], where the dynamics of ammonia adsorption-desorption

Page 32: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 2. State of the art 17

were studied by transient response experiments. These were subsequently modeled by a

Temkin type rate expression for the adsorption and a surface coverage dependent activa-

tion energy for desorption and first order kinetics for the NO reduction. The rate param-

eters were determined by minimizing the error between experimental data and model fit.

This pioneering work forms the basis of many dynamic models presented later.

While in [42] diffusion in the washcoat was regarded as an important factor, a large por-

tion of the ensuing work was directed at improving the understanding of the chemical

kinetics and neglected such effects. E.g. in [50], a catalytic cycle for the standard SCR

reaction over Vanadia-based catalysts was proposed and the parameters for the elemen-

tary steps were determined from experimental data.

Reaction rates based on a non-activated NH3 adsorption process, Temkin-type desorp-

tion kinetics and Eley-Rideal mechanism for the standard SCR reaction were subse-

quently proposed in [51, 52] to estimate kinetic parameters for adsorption/desorption of

ammonia and the standard SCR reaction on Vanadia; both investigations employed a 1D

model for the channel and did not account for diffusion inside the catalytic layer. In par-

allel, kinetic parameters were determined based on experimental data using first-order

reaction rates for NH3 and NO adsorption and NO reduction including as well ammonia

oxidation [53].

Transient kinetic behavior of the standard SCR reaction on Vanadia was studied in more

detail including the adsorption, desorption and oxidation of ammonia. The mass balance

was modeled by a 1D code for the gas phase without diffusion into the wall [54].

As the next step of improving the kinetic model, the role of NO2 on Vanadia-based cata-

lysts was experimentally investigated and a mechanism for the NO/NO2-SCR reaction,

called fast SCR as eq. (1.5), was proposed [55]. In [25], the transient response of the

system in the presence of NO2 was experimentally studied by means of crushed mono-

liths and new elementary reactions on Vanadia catalysts were proposed to explain the

behavior of the system in the fast SCR reaction. More recently, the catalytic redox cycle

Page 33: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

18 Chapter 2. State of the art

and the reaction network of the underlying chemistry in the standard and fast SCR reac-

tion was clarified as a function of temperature and NO2 feed content [56, 57].

In mobile applications of ammonia SCR [58], operating conditions of converters were

strongly transient and a broad temperature window was covered during regular use. NH3

inhibition effects were seen to be more significant at low temperatures and affected the

dynamic response of the SCR systems. The traditional standard Eley-Rideal rate law for

the SCR reaction [54] did not consider the inhibiting effect of NH3 on the SCR reaction

which leads decreasing of NO to minimum after ammonia shut off because of inhibiting

effect of absorbed ammonia. As shown in [59], this inhibition effect had pronounced

practical consequences and should, therefore, be included in the kinetic model of SCR

systems in vehicles. Under such highly transient conditions also the urea dosage was

highly dynamic and, as a consequence, the concentration of the urea decomposition

product NH3 was seen to change considerably during an NH3 shutdown phase.

In 2005, a transient kinetic of the Standard SCR reaction between NO and NH3 was

studied and the new catalytic kinetics and mechanisms dominating at low temperatures

were introduced over a commercial powdered V2O5/WO3–TiO2 [59]. The NH3 inhibi-

tion effect which was observed during transient experiments at low temperatures (lower

than 250 ºC) was explained by an original dual-site modified redox rate law. Two dif-

ferent types of sites were proposed on the surface, namely redox and acidic sites. The

proposed kinetic equation considers adsorption on both sites, the standard SCR reaction,

reoxidation of reduced redox sites and NH3 spill-over between the sites. A modified re-

dox (MR) rate law for the standard SCR reaction was derived by assuming quasi-

equilibrium conditions for the NO adsorption as well as NH3 spill-over. It was demon-

strated, that the 1D+1D model for the SCR converter employing these novel modified

redox kinetics successfully predicted the transient runs at different scales, including ex-

periments of the full scale SCR in an engine test bench. The MR kinetic model was

validated by comparing its numerical results with data from microreactor experiments

with a pulsed NH3 feed and it was concluded that changing the DeNOx rate equation

Page 34: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 2. State of the art 19

from the Eley-Rideal-based reaction rate [54] to the new MR kinetic model improved

the prediction of fast transient operations of the SCR in vehicles after treatment systems.

The Oxygen influence on SCR over the V2O5/WO3–TiO2 catalyst as well as ammonia

inhibition at low temperatures were specifically addressed in [58]. The reported obser-

vations proposed the existence of an optimal ammonia surface coverage lower than the

coverage at steady state. NH3 inhibition was explained by the rate limiting step of the

Standard SCR reaction, ‘which was adversely affected by NH3 adsorbed onto nearby

acidic sites, due to electronic interactions or possibly direct blocking of the redox sites’.

An increase of the Oxygen content did not change the ammonia inhibition effect in the

transient experiments. The main effect of the higher Oxygen concentration was slightly

improving the average SCR conversion. Variation of the ammonia on the redox site was

evaluated by a kinetic approach instead of fast ammonia spill-over assumption from

[59]. The rate of NH3 spill-over was assumed to be similar to the rate of NH3 build-

up/depletion on acidic sites, which determined the characteristic time for the SCR tran-

sients. Since the SCR mechanism was shown to be governed by the reoxidation of the

reduced redox sites by Oxygen, the kinetics depended on the Oxygen content in the feed

and the model describes well the experimental Oxygen effect. The transient effects of

ammonia inhibition observed at low temperature during the NH3 shutdown phase exhib-

ited characteristic timescales of several minutes, which were comparable to the charac-

teristic times for build-up/depletion of ammonia adsorbed onto the SCR catalyst.

A set of elementary reactions for a NO/NO2/NH3 system on Vanadia catalysts was in-

troduced in [24] to simulate a single channel in a transient 1D+1D code and predict the

transient response of the system. The simulation results revealed that the presence of

NO2 at the SCR inlet significantly increased the NOx conversion efficiency. However,

an appropriate design of the oxidation catalyst was necessary. Additionally, the pres-

ence of NO2 in the inlet gas feed enabled a reduction of the catalyst volume by keeping

the SCR efficiency on the same level. Finally, the monolith converter was modeled by

the same group in 1D+1D transient code considering the newly identified elementary

reactions from [60].

Page 35: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

20 Chapter 2. State of the art

A few years ago metal-exchanged Zeolite catalysts were identified as an interesting al-

ternative for the established Vanadia-based SCR catalysts [61]. Accordingly, the re-

sponse of the SCR system on step changes was compared for Vanadia and Zeolite cata-

lysts by using similar reaction rate expression with different parameters for both sys-

tems [29]. A numerical model was presented to simulate coated and extruded monoliths.

The simulation argued that with equal mass of catalytic active material, the Vanadium

and Zeolite catalyst had a similar NOx conversion efficiency. However, based on the

same volume of active material, NOx reduction with the Vanadium catalyst was more

efficient for the ratios of NO2/NOx below 50%, especially at lower temperatures. This

difference was magnified since the Vanadium catalyst was examined in the extruded

monolith and more catalytic active material was provided than a coated Zeolite catalyst.

ZeoliteOn the other the NOx conversion was more efficient over the Zeolite in optimum

operating conditions at higher temperatures.

Global reaction rates were proposed for NH3 adsorption/desorption/oxidation, NO oxi-

dation, standard and fast SCR reaction to model the system for both catalysts in a

1D+1D transient code in [30]. Analogous to Vanadia-based catalysts, the basic reaction

steps of the NO/NO2/NH3 SCR chemistry were elucidated for Fe-Zeolite catalysts by

transient response experiments. It was established that the mechanistic pathways

demonstrated for V-based catalysts widely apply also for Fe-exchanged Zeolites and a

general mechanism was considered to describe the Fast SCR reaction at low tempera-

ture over V-based and Zeolite catalysts.

In [31], an experimental study of the NH3-NO-NO2 SCR reactivity over a commercial

washcoated Fe-Zeolite catalyst was presented in view of its use in Diesel exhaust

aftertreatment technologies. The effects of the main operating conditions were tested

and experiments showed that Oxygen concentration (2-10% v/v) has no effect on NH3

adsorption and Fast SCR (reaction between NO/NO2 in a 1/1 ratio and NH3), however it

promotes NH3 and NO oxidation as well as Standard SCR and inhibits NO2 decomposi-

tion. Compared to a V-based SCR catalyst, Fe-Zeolites are characterized by a higher

activity in the Standard SCR reaction at intermediate and high temperatures, a generally

Page 36: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 2. State of the art 21

much higher activity in the Fast SCR reactionZeolite and a higher selectivity to N2O in

the presence of excess NO2 as was also shown in [29, 56-59]

Standard and Fast SCR reactions were unified into a single redox approach by applying

a model of dynamic Mars-van Krevelen kinetic for V-based catalysts Diesel[62]. A

global kinetic model was derived for the full NH3-NO-NO2 SCR reacting system by ex-

tending the dual site redox kinetics presented in [58] for the NH3-NO-O2 reacting sub-

system. It was shown that NO2 plays no direct role in the redox cycle [56], but forms

nitrates on the catalyst surface, which were responsible for a faster reoxidation of the

reduced redox sites. Considering the rates of reduction and re-oxidation were equal, the

reduction rate was developed by accounting for the complete set of processes: catalyst

reduction/oxidation, nitrate formation and ammonia spill-over. The resulting kinetic

model expressed the overall reduction rate of NO associated with both the Standard and

Fast SCR reaction.

The activity of a commercial copper-exchanged Zeolite SCR catalyst for heavy and

light duty Diesel engines was investigated by both steady state and transient experi-

ments and the results were compared with Fe-Zeolite catalysts in [28]. A higher ammo-

nia storage capacity was observed for the copper Zeolite than the iron-based catalyst.

The ammonia oxidation and the Standard SCR reaction were more active over copper

Zeolite catalyst and the DeNOx activity of Fe-Zeolite was more sensitive to the NO2

feed content. ZeoliteThe inhibiting effect of ammonia on the Standard SCR reaction

was stronger over the iron Zeolite.

Finally, NOx reduction in ammonia SCR was simulated for a combined NOx storage and

NH3 SCR catalytic system in [63] by using the dual site modified redox rate reaction for

Standard SCR [58]. It was revealed, that the transient NH3 inhibition effects over the

Fe-ZSM-5 catalysts were correctly predicted by a finite spill-over reaction rate. Alt-

hough the equilibrium assumption for NH3 spill-over was acceptable for many other

SCR catalysts (such as Cu-ZSM-5), this simplification was not satisfactory for iron Zeo-

lite.

Page 37: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

22 Chapter 2. State of the art

2.2 Conclusion

The Selective Catalytic Reduction (SCR) of NOx by ammonia is a promising technique

for the abatement of NOx from the exhaust of Diesel engines. The design of an SCR

system is a complex process involving optimization of size, density and chemical com-

position of the SCR monolith as well as the ammonia dosing system. In automotive ap-

plications, the transient conditions during acceleration, such as the heat-up of the cata-

lytic converter, pose additional challenges. In view of shortening the development cycle

and reducing the costs, numerical simulation is a powerful method to investigate the

above processes.

A variety of modeling approaches are in use today depending on the process of interest.

Regarding the fluid flow, simulation is mostly often performed for a single channel of a

honeycomb SCR converter. The simulation can be based on a full 3D model, in many

cases 2D or 1D lumped parameter models are used for boundary conditions at the cata-

lytic wall. While diffusion within the wall is not considered in some early models, more

recent approaches include simulation of the wall in 1D or 2D and the necessity to ac-

count for diffusional limitations in the washcoat has been clearly demonstrated, e.g. [38,

44]. Modeling of the surface kinetics plays an important role since a number of inter-

mediate and main species and reactions must be considered in a SCR system. The selec-

tion of the most important and stable species and the way different reactions are consid-

ered has a significant impact on the results and the simplicity of the system modeling.

Table 2.1 reviews the literatures investigating the simulation of the system and SCR ki-

netics, as explained in the last sections. Simulation of the channel is mentioned as one

dimensional (1D), two dimensional (1D), three dimensional (1D), Navier-Stokes (NS),

or boundary layer (BL). In simulation of the catalytic layer (wall), it is assigned whether

intra-porous diffusion is modeled (1D) or neglected, where the chemical reactions rates

are considered as a source term at the wall (source). It is as well specified whether the

governing equations are considered in steady or transient cases. Different proposed ki-

netics for SCR reactions is mentioned and the type of the catalyst is named as Vanadi-

Page 38: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 2. State of the art 23

um- (Van) or Zeolite- (Ze) based. The literatures that experimentally studied the re-

sponse of the system are marked (EXP) and the new aspects of each article are also

briefly mentioned. Some articles are reported which are not specifically about SCR

converter but present useful comparisons of different models in the monolith converters.

As the current state of the art, 1D+1D model are a good choice in most cases, consider-

ing not only the convective mass transfer in the channel but also the intra-porous diffu-

sion within the washcoat layer on the channel walls [36, 38, 41]. The most significant

progress in the simulation of the SCR system behavior is achieved by improving the

modeling of the surface kinetics. Various surface kinetics and species have been imple-

mented in new models explaining the detailed mechanisms. The effect of ammonia ad-

sorption/desorption on transient response of the system is considered [42, 51, 52] and

the inhibition effect of ammonia is introduced which is an important and determinant

fact in the SCR system and affects the performance [64]. Reaction rate of the Standard

SCR is improved by new kinetics to capture the inhibition effect [58, 59] and Fast SCR

reaction is also considered in the new kinetics [25, 29]. The new literature proposed

more detailed reactions of minor species to improve the prediction of the model in dif-

ferent conditions [30, 62]. Ze-based catalyst is introduced for the SCR system and com-

pared by V-based catalyst [28, 61]. The SCR kinetics is also studied on Ze-catalyst and

improved to satisfy the behavior of the system [31, 63].

Page 39: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Table 2.1. Sate of the art

Time Channel Wall Catalyst Reaction (Kinetics) Aspect Article

Steady1 1D2 Source3 - General Pioneer work [33]

Exp4 Exp Exp Van5 Detailed SCR mechanism Pioneer work in kinetics [49]

Steady 1, 26, 3D7 Source Van Standard SCR (Rideal) Compare models [36]

Steady 1D 1D8 Van Standard SCR (Rideal) Diffusional limit [37]

Steady 1, 2, 3D Source - Catalytic combustor Compare models [41]

Transient9 1D 1D Van NH3 ads-des Kinetics [42]

Exp Exp Exp Van SCR catalytic cycle Kinetics [50]

Steady 3D Source - Catalytic combustor Triangular channel [45]

Steady 1D 1D Van Standard SCR (Rideal) Calibration of properties

and parameters [43, 53]

Page 40: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Time Channel Wall Catalyst Reaction (Kinetics) Aspect Article

Transient 1D Source Van NH3 ads-des,

Standard SCR (Rideal) Kinetics [51, 52]

Steady 1D 1D Van Standard SCR, NH3 oxidation

Kinetic parameter [48]

Steady NS10, BL11,

1D Source - Catalytic combustor Compare models [47]

Exp Exp Exp Van Fast SCR NO/NO2 SCR [55]

Transient 1D Source Van NH3 ads-des-ox,

standard SCR (Rideal) Kinetics [54]

Transient 1D 1D Van NH3 ads-des-ox,

standard SCR Kinetics

(dual-site, modified redox) [38, 59]

Transient 1D Source Van NH3 ads-des-ox,

standard SCR Ammonia inhibition [58]

Exp Exp Exp Van Fast SCR NH3-NO/NO2 chemistry [25]

Page 41: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Time Channel Wall Catalyst Reaction (Kinetics) Aspect Article

Exp Exp Exp Van Standard and Fast SCR Redox catalytic cycle [56]

Exp Exp Exp Van NH3 ads-des-ox, standard

and Fast SCR, NH4NO3 reactions NH3-NO/NO2 reaction network [57]

Transient 1D 1D Van NH3 ads-des-ox, standard

and Fast SCR, NH4NO3 reactions Simulation [60]

Transient 1D 1D Van, Ze12 NH3 ads-des-ox,

standard and Fast SCR Different catalysts [29]

Exp Exp Exp Ze Fast SCR NH3-NO/NO2 chemistry

on Fe-Ze [30]

Exp Exp Exp Ze NH3-NO/NO2 chemistry Study and compare catalyst [31]

Transient 1D Source Van Standard and Fast SCR Overall DeNOx kinetics [62]

Transient 1D 1D Ze NH3 ads-des-ox, standard

and Fast SCR, NH4NO3 reactions Non-equilibrium spill-over [63]

Steady NS, BL, 1D 1D, Source - Three-way catalyst Compare models [44]

Page 42: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Abbreviations:

1 Steady state simulation

2 Channel is simulated in one longitudinal direction

3 Wall is simulated as a source term for the channel

4 Experimental study

5 Vanadia-based catalyst

6,7 Channel is simulated in two and three directions

8 Wall is simulated in one perpendicular direction

9 Transient simulation

10 Channel is simulated based on Navier-Stokes equations

11 Channel is simulated based on Boundary-Layer equations

12 Zeolite-based catalyst

Page 43: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären
Page 44: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 3. Methodology

This chapter presents equations governing the flow, species and energy in channels and

catalytic layers of the converter. Different models for the channel and the catalytic wall

are mentioned and the related boundary conditions are explained. Catalytic surface reac-

tions of the SCR system are introduced and different proposed kinetics is reported. Fi-

nally the selected model is explained to simulate the response of the system and the nu-

merical method is described to solve the problem.

3.1 Modeling of the monolith converter

A typical SCR converter has a large number of parallel channels in a honeycomb mono-

lith structure with catalyst washcoated walls. To simulate the behaviour of the SCR

process one representative channel is selected. The model accounts for the convective

mass transfer through the channel, the diffusive mass transfer and the chemical reac-

tions within the catalytic layer coupled together via the mass transfer boundary condi-

tions at the wall.

3.1.1 The converter channel

The transient convective mass transfer inside the channel is described by eq. (3.1) in 3D

Cartesian coordinates for square and triangle channels [41].

2 2

, 2 2i i i i

g iC C C Cu Dt x z y

∂ ∂ ∂ ∂= − + + ∂ ∂ ∂ ∂

(3.1)

Mass transfer is considered in 2D cylindrical coordinates for circular channels accord-

ing to eq. (3.2) [41].

,1i i i

g iC C Cu D rt x r r r

∂ ∂ ∂∂ = − + ∂ ∂ ∂ ∂ (3.2)

Page 45: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

30 Chapter 3. Methodology

In these models species diffusion in radial direction, (z, y, r), is considered while axial

diffusion is neglected along the channel. It was reported by [36, 65] that axial molecular

diffusion effects are important only at low values of Peclet number showing the ratio of

the convective mass transfer rate to the rate of the diffusive mass transfer,

Pe = (u Dh/Dg) less than 100.

Beside the 3D and 2D models for simulation of the channel, lumped parameter model is

a promising tool to analyze and design the SCR reactors, concluded by [36] as men-

tioned in section 2.1.1. The lumped parameter model considers the average value for

velocity and concentration over the channel cross section in the gas phase and a periph-

eral average of the concentration over the catalytic wall. According to this, 1D mass

transfer equation in the channel is described by eq. (3.3) and modelled by means of a

hyperbolic equation for a step flow into the channel.

( ),

interface

4∂ ∂= − − −

∂ ∂

m ii ii i

h

kC Cu C Ct x D

(3.3)

where u denotes the average velocity and Ci the species concentrations in each section

of the channel. The last term in eq. (3.3) describes the mass transfer at the boundary

condition between channel and wall.

Similarly the transient convective heat transfer inside the channel is represented in 1D

as eq. (3.4).

( )interface

( ) ( ) 4g gg p g p h

h

c T c uT k T Tt x D

ρ ρ∂ ∂= − − −

∂ ∂

(3.4)

3.1.2 Catalytic washcoat layer

The physical and chemical processes within the catalytic layer comprise several steps

including diffusion and surface reactions. The gaseous flow inside the channel contains

the species O2, NO, NO2, NH3, N2, H2O and N2O. The gaseous species diffuse into the

Page 46: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 3. Methodology 31

washcoat layer and ammonia inside the wall pores adsorbs strongly on the catalytic sur-

face. The adsorbed ammonia reacts with NOx and O2, which diffuse into the pores. As a

result water vapor and Nitrogen are produced and diffuse back into the channel.

In a simple model of the washcoat layer, rate of catalytic reactions are calculated based

on the species concentration inside the channel adjacent to the wall and there is no dis-

tribution of the adsorbed species within the layer [41]. In this model, the formation rate

of species is considered at the wall boundary condition as explained in section 3.1.3.

Intra-porous diffusion limits ammonia adsorption on the catalytic surface within the

washcoated layer and consequently influences NOx reduction as is clearly demonstrated

e.g. in [38, 44]. The diffusion within the catalytic layer is described by the following

equation:

2

2

∂ ∂= +

∂ ∂

i ieff i

C CD rt y

(3.5)

where iC

denotes the concentration of the species NO, NO2 and NH3 inside the pores

and ri shows the formation rate of species i due to the surface reactions. Distribution of

species is considered perpendicular to the wall and diffusion in the directions parallel to

the wall is neglected [46].

Heat transfer in the wall comprises the conduction along the channel balanced by heat

transfer boundary condition at the wall and heat of the reactions within the catalytic lay-

er, eq. (3.6). ( )

T x represents the wall temperature distribution in longitudinal direction

of the channel. There is no temperature gradient in normal direction of the wall since the

thermal conductivity is sufficiently high and the washcoat layer is very thin [66].

( )2

2

( )

( ) ( )effeff p h h

eff h H rh h

c T D DTk k T T q qt x D D

ρ

δ δ δ δ

∂ ∂= + − + +

∂ ∂ + +

(3.6)

Page 47: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

32 Chapter 3. Methodology

where the effective parameters (keff, ρeff, cp,eff) are calculated for the solid phase, includ-

ing the monolith substrate, the washcoat layer and the catalyst. qH is external heat

source at the wall over peripheral area of the cells and qr shows the volumetric heat of

surface reactions.

3.1.3 Boundary condition at the catalytic wall

Mass transfer inside the channel and within the wall are coupled via a mass transfer

boundary condition at the wall in normal direction n, which is derived by gas diffusion

inside the channel and intra-porous diffusion within the wall as generally described by

eq. (3.7) [46]. Dg,i is a property of gas showing bimolecular diffusion of the species and

Deff,i presents effective diffusivity of the species into the washcoat layer which is a

property of the layer and discussed later in section 5.3.1.

, ,interface,g interface,w

i ig i eff i

C CD Dn n

∂ ∂=

∂ ∂

(3.7)

This general boundary condition is usually simplified by assumptions for the wall and

flow inside the channel. A simplified model for the wall which includes surface reac-

tions but neglects the intra-porous diffusion [41] and the mass transfer at the wall is de-

scribed by eq. (3.8), where species diffusion inside the channel is equal to consump-

tion/production of species on the catalytic surface. r shows the rate of formation of each

species and δ is the thickness of the catalytic layer.

,interface

δ∂= − ⋅

∂i

g i iCD rn

(3.8)

According the lumped parameter model for simulation of the channel [36], mass trans-

fer boundary condition at the wall is simplified to eq. (3.9) where km,i is the convective

mass transfer coefficient of the channel calculated from km,i = (Sh Dg,i)/x. A technical

approximation is assuming Sherwood number equal to Nusselt number from the solu-

Page 48: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 3. Methodology 33

tion of the Graetz-Nusselt problem for constant wall temperature which meant neglect-

ing the influence of the kinetics in the dimensionless mass transfer coefficient.

, interface( ) δ− = − ⋅

m i i i ik C C r (3.9)

Sherwood number is calculated by eq. (3.10) from Graetz-Nusselt analogy for a devel-

oping laminar flow in a square duct according to the expression proposed by [36, 38]

and using the same coefficients.

3 0.488 2+6.874( 10 ) exp( 57.2 ) where ( . ) / ( . )−= × − =s s s g hSh Nu x x x x D u D (3.10)

Dh is hydraulic diameter of the converter channel with square cross section. The chan-

nels in the monolith structure are typically square ducts and the deformation of the cross

section due to washcoating is neglected since the washcoat layer is very thin. 2D analy-

sis showed that Sherwood number decreased rapidly along the entry region of the chan-

nel until an asymptotic value, depending on both geometry and gas properties, is

reached [36].

The analogies between the SCR and Graetz-Nusselt problems, which governed heat

transfer to a fluid in laminar flow in a duct with constant wall temperature, were inves-

tigated in [36] and shown that these two problems were similar except for the boundary

condition at the wall. Sherwood number played the similar role as the Nusselt number

whereas it was the difference where there was no chemical phenomenon in Graetz-

Nusselt problem. The exact analogies between these two problems were in two limits.

The first is infinitely fast kinetics (Damkohler, Da→∞) which corresponded to a purely

physical regime at the wall. This condition was the same as the Graetz-Nusselt problem

with constant wall temperature. The second was infinitely slow kinetics (Da→0) which

implied a purely chemical regime. This condition was matching the Graetz-Nusselt

problem with constant heat flux at the wall.

Page 49: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

34 Chapter 3. Methodology

Intra-porous diffusion within the washcoat layer should be considered in the model

since the diffusional resistance limits the SCR performance [38, 44] as discussed in sec-

tion 5.2. Therefore mass transfer boundary condition at the wall is described by eq.

(3.11) according to the lumped parameter model in the channel and accounting intra-

porous diffusion within the wall.

, interfaceinterface

( ) im i i i eff

Ck C C Dy

∂− =

(3.11)

For non-isothermal conditions, the model accounts for the convective heat transfer

through the channel, the conductive heat transfer in the wall along the channel and the

heat of chemical reactions within the catalytic layer coupled together via the equal heat

transfers boundary conditions at the wall,

interface( )= −

wall hq k T T (3.12)

where kh is the convective heat transfer coefficient of the channel calculated from

kh = (Nu kg)/Dh based on Nusselt number proposed by [41] for a laminar flow in a

square duct, eq. (3.13). Peclet number (Pe) in thermal diffusion is Pe = (u Dh/α) where

α represents the thermal diffusivity (α = kg/ρg cp,g).

3 0.51742.977+6.845( 10 ) exp( 42.49 ) where / ( . )− ′ ′= × − =s s s hNu x x x x D Pe (3.13)

3.2 Kinetics of catalytic heterogeneous reactions

To calculate the formation rate of species inside the pores, appropriate chemical kinetics

are needed for the catalyst and the operating conditions, which can describe the behav-

iour of the system. The catalytic surface reactions consist of three groups. One group

includes the reactions relating to adsorption and desorption of ammonia. Group two

consists of actual SCR reactions such as standard, fast and minor detailed reactions ex-

plaining the reduction of NOx. The third group is side-reactions like NO and NH3 oxida-

Page 50: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 3. Methodology 35

tion. The kinetics of each group of reactions on Fe-Zeolite catalyst are discussed in the

next sections.

3.2.1 Ammonia adsorption/desorption/spill-over

The first step in the SCR reaction is the adsorption/desorption of ammonia onto the

catalyst, eq. (3.14). NH3 is adsorbed on the acidic site of the catalyst and provides a res-

ervoir for NH3 storage/reaction [51].

NH3 ↔ NH3* (3.14)

A non-activated NH3 adsorption process and a Temkin-type NH3 desorption with linear

coverage dependent activation energy are assumed for the reaction rates, eqs. (3.15) and

(3.16), respectively, according to the basic study by [42].

3

0 (1 )ω∗= −

ads ads NHrr k C (3.15)

0 exp (1 )γω ω∗ ∗ = − − des

des desErr kRT

(3.16)

where ω* refers to ammonia surface coverage in the acidic site.

In order to take into account ammonia inhibition effect, an alternative rate expression

was implemented by [59] assuming two different types of sites on the surface of the cat-

alyst. There are one redox site for O2 and NO adsorption/activation and one acidic site

for NH3 adsorption. In the presence of Oxygen, there is an ammonia spill-over reaction

involving adjacent acidic and redox sites, as eq. (3.17) which shows the replacement of

adsorbed ammonia between two the sites [59].

NH3* (acidic site) ↔ NH3

• (redox site) (3.17)

The corresponding rate for the ammonia spill-over reaction can be expressed as eq.

(3.18) following [58].

Page 51: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

36 Chapter 3. Methodology

( ) ( )01

1ω ω

ω ω• ∗

∗ •− −

− = − −

s o s oLH

rr kK

(3.18)

ω• refers to ammonia surface coverage in the redox site and KLH represents the equilib-

rium constant of reaction (3.17).

Accumulation and depletion of the absorbed ammonia is modelled by eq. (3.19) for both

acidic and redox sites on the catalytic surface [63].

3 , *( ), ( )ω∂

Ω = = •∂

i

ii

NHr i acidic redox

t (3.19)

where ω and Ω denote the ammonia surface coverage on acidic and redox sites and the

ammonia storage capacity, respectively. The capacity to store ammonia in the Zeolite

structure (acidic site) is assumed to be about 450 times the number of reactive (redox)

sites following consideration put forward in [63] for a similar catalyst. r represents the

formation rate of ammonia on both surface sites.

Since the rate of formation of adsorbed ammonia (NH3* and NH3

•) depends on the con-

centration of species within the catalyst layer, eq. (3.19) is based on the void volume of

the washcoat layer. Therefore, the adsorption capacity, Ω, is also needed per void vol-

ume, which depends on the washcoat properties such as porosity ε in eq. (3.20)

3 3

3 3 3 3

1 , ,ε

Ω = Ω Α Β = ∗ •

i i cat cell washcoat

cvoid cat cell washcoat void

kg m mmol mol im kg m m m

(3.20)

where parameters A and B are converter characteristics showing the active mass of cata-

lyst used in the converter and volume of washcoat layer in each cell, respectively. Ωc is

the property of the catalyst showing the capacity for the adsorbed moles.

Page 52: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 3. Methodology 37

3.2.2 Standard and fast SCR

The catalytic reduction of NOx by ammonia is governed by two main reactions, standard

and fast SCR as described in section 1.2.2. Three types of kinetics are assessed to model

the kinetics of standard and fast SCR and the results are improved by using more de-

tailed kinetics as are discussed in the following chapters.

In the first model, developed by [59], the reaction rate of standard SCR, eq. (3.21) is

derived by a simplification of the expression proposed by [58] for quasi-equilibrium

conditions on the surface, eq. (3.21).

3 2

3

3

3

1 1exp473 0.02

11

βω

ωω

= − − + −

NO NH Ostrdstrd strd

NHNH

NH

C pErr kR T

K (3.21)

In the simplified reaction rate expression from [59] the kinetic dependence on O2 is in-

corporated into an empirical power law term. This simplification decreases the depend-

ence of the rate on the surface coverage. Therefore the reaction rate of the standard SCR

is not simplified in the second model, eq. (3.22), [58] with the same basis as the rate

used in the first model.

3

3 3

3 2

3 2

0.25

1 1exp473

1 11

ω

ω ωω

∗ ∗

= − − + + −

NO NHstrdstrd strd

NH NO NHNH O

NH O

CErr kR T C

K Kp

(3.22)

The reaction rate of the fast SCR for both models one and two are the same as proposed

by [29]:

2

3

2

1 1exp473

ωκ

∗ = − − +

NOfastfast fast NO NH

NO

CErr k C

R T C (3.23)

Page 53: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

38 Chapter 3. Methodology

In the third model, the rate law for the Standard SCR reaction is based on a dual-site

modified redox (MR) rate law, eq. (3.24), accompanied by applying the finite reaction

rate for ammonia spill-over [58], eq. (3.18). Following considerations from [63], NH3

can block the active redox sites, which corresponds to the observed inhibition of the

Standard SCR reaction. As explained in section 2.1.2, only a finite spill-over reaction

rate is able to reproduce the transient NH3 inhibition effects on the Fe-ZSM-5 catalysts.

( )( )2

011 1exp

473 1

ω ω

ω

∗ •

−− = − +

NOstrdstrd strd

O NO

CErr kR T K C

(3.24)

The presence of NO2 leads to a higher NOx conversion rate due to the Fast SCR reac-

tion. The reaction rate of the fast SCR in the third model is calculated by [63] as eq.

(3.25). Comparing the reaction rates of fast SCR in the three models, eqs. (3.23) and

(3.25), the rate of model three has a simplification for parameter κ, which is merged into

the rate coefficient, since the value of NO2 concentration is negligible compared to κ.

2

0 1 1exp473

ω∗− = −

fastfast fast NO NO

Err k C C

R T (3.25)

3.2.3 NH3 oxidation

The oxidation of ammonia is observed at high temperature, eq. (3.26) [64].

4 NH3* + 3 O2 → 2 N2 + 6 H2O (3.26)

The rate of the ammonia oxidation is accounted by the simple Arrhenius expression,

eq. (3.27) [64].

0 1 1exp473

ω∗− = − ox

ox oxErr kR T

(3.27)

Page 54: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 3. Methodology 39

3.2.4 NO oxidation

The thermodynamic equilibrium of NO and NO2 is also modelled to cover the small NO

oxidation, as eq. (3.28), activity found on Fe-Zeolites [31, 67].

2 NO + O2 → 2 NO2 (3.28)

The reaction rate of NO oxidation, eq. (3.29) based on [29, 63], includes the semi-

equilibrium constant eqK

which is different from the equilibrium constant in terms of

the Gibbs energy change, since the NO oxidation is found to be far from equilibrium (as

discussed in section 5.3.2). The semi-equilibrium constant is defined to comprise the

kinetic effects and mass transfer limits of the system.

( )2

2

0 1 1exp , exp473

− − − = − − =

NONONO NO NO O eq

eq

C a bTErr k C C KR T K RT

(3.29)

The two parameters a and b are determined in such a way that eqK

satisfies the steady

state concentrations of NO and NO2 for different conditions.

3.2.5 NO2 and NH3 reactions

Three reactions between NO2 and NH3 are considered, involving in the SCR system for

specific conditions. NO2 was allowed in the model to react directly with NH3 at high

temperatures [63], eq. (3.30).

2 NH3* + 3/2 NO2 → 7/4 N2 + 3 H2O (3.30)

The rate of the direct reduction of NO2 is described as eq. (3.31) following [63].

2

2 2 2

0 1 1exp473

ω∗− = −

NONO NO NO

Err k C

R T (3.31)

Page 55: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

40 Chapter 3. Methodology

In addition, high NO2/NOx ratios may also result in the formation of NH4NO3 and N2O,

eqs. (3.32) and (3.33) respectively. It is assumed that NH4NO3 decomposes again,

which is true only for temperatures higher than 200 ºC [63].

2 NH3* + 2 NO2 → NH4NO3 + N2 + H2O (3.32)

2 NH3* + 2 NO2 → N2O + N2 + 3 H2O (3.33)

The formation rate of ammonium-nitrate and nitrous-oxide are calculated according to

eqs. (3.34) and (3.35) respectively, as proposed in [63].

3

3 3 2

0 21 1exp473

ω∗− = −

NONO NO NO

Err k C

R T (3.34)

2

2 2 2

0 1 1exp473

ω∗− = −

N ON O N O NO

Err k C

R T (3.35)

3.3 Heat of reactions

In order to develop a kinetic model for ammonia SCR capable of describing transient

effects, it is vital to obtain an accurate description of the storage behavior of the cata-

lyst. The adsorption heat is described as eq. (3.36), which is a key parameter for model-

ing the exothermic ammonia storage process [68]. This relation is taken from a similar

expression for the coverage dependent activation energy of ammonia desorption, eq.

(3.16)

(1 )ads desH E γω∗∆ = − − (3.36)

To solve the heat balance in the system, the enthalpy of the ammonia adsorption is con-

sidered, however the reaction heat associated with NH3-SCR is very low [69].

Page 56: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 3. Methodology 41

3.4 Transient 1D+1D model including heterogeneous reactions

As the conclusion of the presented models for mass and heat transfer simulation and ki-

netics of chemical reactions, the transient 1D+1D model is selected to simulate the SCR

system. 1D+1D model, including the lumped parameter model for the channel and intra-

porous diffusion within the wall, is concluded as a promising method to predict the re-

sponse of the SCR system in transient operation [36, 38]. The governing equations are

summarized bellow for convective mass transfer inside the channel eq. (3.3), species

diffusion within the wall, eq. (3.5) and boundary condition at the wall as eq. (3.11).

( ),

interface

4∂ ∂= − − −

∂ ∂

m ii ii i

h

kC Cu C Ct x D

(3.3)

2

2

∂ ∂= +

∂ ∂

i ieff i

C CD rt y

(3.5)

, interfaceinterface

( ) im i i i eff

Ck C C Dy

∂− =

(3.11)

Ten surface reactions are taken into account for the Fe-Zeolite catalyst according to [63]

which are summarised in Table 3.1, and the reaction rates are shown as well. Rates of

standard and fast SCR reactions corresponding to three models presented in section

3.2.2 are mentioned here, given by r3 [63], r3′ [58], r3″ [59] for standard SCR and r4

[63], r4′ [29] for fast SCR, and discussed later in sections 5.3.3 and 5.3.4.

3.5 Numerical method

The presented 1D+1D model including the SCR chemical kinetics is employed to simu-

late one representative single channel of the monolith structure, as explained in section

3.4. FORTRAN compiler is selected to execute the own developed code, which can be

well connected to the commercial StarCD software to predict the response of the entire

SCR system. The 1D+1D code solves iteratively the mass transfer inside the system.

The convective mass transfer is solved in one dimension along the channel. Coupled,

Page 57: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

42 Chapter 3. Methodology

the diffusive mass transfer within the wall is calculated in one dimension perpendicular

to the wall according to the channel concentration in each point. Rate of surface reac-

tions in each depth of the wall are calculated based on the species concentrations at that

point.

Finite differences are employed to discretize the spatial domains of both the washcoat

and the channel. The resolution requirements are checked to ensure grid independent

results for nx cells along the channel and nx.ny cells for the wall as discussed later in

section 5.1. An upwind scheme is employed for the convective mass transfer through

the channel to solve the hyperbolic transient first order differential equation. Species

diffusion within the layer is solved by centred differences scheme for the transient sec-

ond order differential equation. Time integration is performed by means of an first order

explicit scheme to capture the inflow steps into the channel as a marching problem with

time steps smaller than the time steps according to CFL stability condition [70]. Diffu-

sion problem is integrated implicitly with small time steps to cover the slow diffusion

process.

Overall time step is selected equal or smaller than CFL time step, ∆tCFL = 0.5 ( ∆x / u ),

to satisfy the slow diffusion. Since chemical reactions are fast at high temperature, very

small time step (∆t = 1/5 ∆tCFL) is needed to fulfil the coupling between diffusion and

reaction. However for low temperature, a larger time step (∆t = 1/5 ∆tCFL) is sufficient

to capture the physical phenomena. The time step is adjusted at each temperature and

selected small enough to guaranty a robust time splitting numerical solution. Hereby, it

is ensured that the robust solution is independent of the time scale.

A time marching numerical algorithm solves channel and wall separately in an iterative

procedure for each time step because of the complexity due to the chemical reactions.

The algorithm is begun by solving the channel with initial conditions, first assumptions

and species concentrations within the wall from the last time step. In the first time step,

they are set according to the initial conditions of the process and the system is filled by

the air. Then the wall is solved by the new data for the channel while initial conditions

Page 58: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 3. Methodology 43

and first assumptions are taken from the last time step. Having the new concentrations

in the wall, the gas species concentration inside the channel are calculated again for the

same time step and compared by the results calculated before. This iterative procedure

is repeated until the same result is obtained therefore the coupled mass transfer bound-

ary condition for the channel and the wall is satisfied. Then the time step is solved and

the algorithm goes to the next time step.

Page 59: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Table 3.1. Surface reactions and their rates

(1) Ammonia adsorption NH3 → NH3*

3

01 1 (1 )NHr k C ω∗= −

(3.15)

(2) Ammonia desorption NH3 ← NH3* 0 2

2 2 exp (1 )Er kRT

γω ω∗ ∗ = − − (3.16)

(3) Standard SCR1 4 NH3* + 4 NO + O2 → 4 N2 + 6 H2O

3 2

3

3

3

3 3 0.021

1

NO NH O

NHNH

NH

C pr k

K

βω

ωω

′′ ′′= + −

(3.21)

3

3 3

3 2

3 2

3 3

0.251 11

NO NH

NH NO NHNH O

NH O

Cr k

CK K

p

ω

ω ωω

∗ ∗

′ ′=

+ + −

(3.22)

( )( )2

3 3

1

1NO

O NO

Cr k

K C

ω ω

ω

∗ •

−=

+

(3.24)

(4) Fast SCR2 4 NH3* + 2 NO + 2 NO2 → 4 N2 + 6 H2O

2

3

2

4 4NO

NO NHNO

Cr k C

κ∗′ ′=

+

(3.23)

24 4 NO NOr k C C ω∗=

(3.25)

Page 60: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

(5) Ammonia spill-over NH3* ↔ NH3

• ( ) ( )05 5

11

ω ωω ω

• ∗∗ •

− = − − LH

r kK

(3.18)

(6) NO oxidation 2 NO + O2 → 2 NO2 ( )2

26 6 , expNONO O eq

eq

C a bTr k C C K

K RT − −

= − =

(3.29)

(7) Ammonia oxidation 4 NH3* + 3 O2 → 2 N2 + 6 H2O 7 7r k ω∗= (3.27)

(8) Direct NO2 reduction 2 NH3* + 3/2 NO2 → 7/4 N2 + 3 H2O

28 8 NOr k C ω∗=

(3.31)

(9) Ammonium nitrate formation 2 NH3* + 2 NO2 → NH4NO3 + N2 + H2O

2

29 9 NOr k C ω∗=

(3.34)

(10) Nitrous oxide formation 2 NH3* + 2 NO2 → N2O + N2 + 3 H2O

210 10 NOr k C ω∗=

(3.35)

1 Reaction rates of standard SCR for the three models are represented by r3 [63], r3′ [58] and r3″ [59].

2 Reaction rates of fast SCR are represented by r4 [63] and r4′ [29].

where 0 1 1exp473

ii i

Ek kR T

= − − (3.37)

Page 61: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären
Page 62: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 4. Experiment

The catalyst used in the experiments is a commercially available Fe-BEA Zeolite for

SCR applications. The powder Fe-BEA material is coated at the Paul Scherrer Institute

on a standard 400 cpsi cordierite monolith from Corning Incorporated (catalyst loading

= 135 g/L) with wall thickness 0.165 mm and calcined at 550 ºC for 4 h prior to testing.

The parameterization experiments were designed by Martin Elsener and Dr. Oliver

Kröcher in Paul Scherrer Institute where the catalyst coating and experiments were done

by Martin Elsener.

For the parameterization of the catalyst model, the catalyst performance has to be meas-

ured in dependency of the parameters NOx,in, NO2,in/NOx,in, α = NH3,in/NOx,in, tempera-

ture and GHSV. For that purpose three small monolith samples with a volume of

6.4 cm3, 8.9 and 13.3 cm3 are cut out of the brick and tested with model Diesel exhaust

gas in the laboratory test apparatus described in [71].

Figure 4.1. Laboratory test apparatus [72], 1: mass flow controller,

2: water supply, 3: reactor, 4: gas cell, 5: analysis software.

Different units of the laboratory test apparatus are shown in Figure 4.1. Inlet species are

set by the mass flow controller in no. 1. Water is supplied into the inlet pipe at no. 2 to

provide the vapor in the feed. The catalytic converter is placed in no. 3 and connected to

the gas cell to control the outlet species at no. 4. The measured data is sent to the com-

Page 63: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

48 Chapter 4. Experiment

puter station in no.5 and analyzed by the software. The connecting pipe and the reactor

are heated and isolated to provide the isotherm condition for the system.

Figure 4.2. Flow and control diagram of the test bench [72], 1: water reservoir, 2: water evaporator,

3: reactor, 4: catalytic converter, 5: filter, 6: flow meter, 7: diaphragm pump, 8: gas cell,

FIC: mass flow controller, TIC: temperature controller, TI: temperature indicator

Figure 4.2 is a schematic of the setup and illustrates the flow and control diagrams. The

gas mixtures enter to the reactor in no. 3 to be heated up to the test temperature which is

controlled in upstream and downstream of the converter. The SCR reduction takes place

in catalytic converter no. 4 and the outlet flow passes through the filter before species

detector at no. 8.

Page 64: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 4. Experiment 49

The characterization experiments are designed such that the catalyst behavior is covered

over a practice-relevant parameter space with a minimum number of experiments:

NOx,in is kept constant at 500 ppm in most of the experiments, the NO2,in/NOx,in ratio is

switched between 0%, 25%, 50% and 75%, the temperature is varied between 200 ºC

and 450 ºC in 50 ºC steps and the GHSV is adjusted to 30,000 h-1, 50,000 h-1, and

70000 h-1 which is calculated at standard pressure and temperature showing the volu-

metric flow rate per volume of the converter. αOD means the optimum dosage ratio of

NH3,in/NOx,in under practice-relevant conditions that results in a slip of 10 ppm ammo-

nia downstream of the catalyst. This ratio, αOD, significantly differs from the ideal 1:1

ratio for NO-SCR (standard SCR), for which it is determined at each temperature and

then varied stepwise. There are 10% Oxygen and 5% water vapor in the feed and the

rest is Nitrogen.

Figure 4.3. Transient operation and steady state measured at GHSV = 50,000 h-1, 200 ºC, 500 ppm NOx, NO/NOx = 100%, NH3/NOx = (0.8, 1.0, 1.2) times of the optimum dosage (αOD for 10 ppm am-

monia slip), Symbols: experimental data, dashed line: inlet, red: NO, blue: NO2, green: NH3.

0

50

100

150

200

250

300

200

250

300

350

400

450

500

2000 3000 4000 5000 6000 7000 8000

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

NOin

NH3,in

NOout

NO2,out

NH3,out

Page 65: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

50 Chapter 4. Experiment

In a first set of step-response experiments the NO-SCR performance and NH3 oxidation

are measured at given temperature and GHSV by switching NH3 on and off at 0.8 αOD,

1.0 αOD and 1.2 αOD. Moreover, NH3 is switched on and off without NOx in the feed in

order to determine the dynamics of the NH3 storage and release. As equilibration is

awaited after the concentration steps, the steady-state performance of the catalyst is in-

cluded in the recorded data, as represented in Figure 4.3. The NO/NO2-SCR perfor-

mance is measured with the same procedure.

Figure 4.4. Highly transient operation and steady state measured at GHSV = 50,000 h-1, 200 ºC, 500

ppm NOx, NO/NOx = 75%, NH3/NOx = (0.8, 1.0, 1.2) times of the optimum dosage (αOD for 10 ppm

ammonia slip), Symbols: experimental data, dashed line: inlet, red: NO, blue: NO2, green: NH3.

The dynamic behavior of the catalyst during NO- and NO/NO2-SCR is tested at

0.8 αOD, 1.0 αOD and 1.2 αOD with 30 s, 60 s, 120 s and 240 s NH3 pulses, separated by

pauses of the same length, as shown in Figure 4.4. For correction of the results for the

0

100

200

300

400

500

600

700

800

900

0

50

100

150

200

250

300

350

400

6900 7900 8900 9900 10900 11900 12900

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

NOin

NH3,inNOout

NO2,out

NH3,out

NO2,in

Page 66: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 4. Experiment 51

influence of the test apparatus, a reference experiment without catalyst is carried out

under NO- and NO/NO2-SCR conditions.

Finally temperature step experiments are carried out for selected conditions, in which

the temperature is ramped from 200 ºC to 450 ºC at 75 ºC/min.

Page 67: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären
Page 68: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 5. Calibration procedure and results

Selective catalytic reduction of NOx by ammonia is numerically simulated and the pre-

dictions are compared with the experiments to verify the model and the calibrated pa-

rameters. This chapter presents the performance of the system in different cases and the

influence of the operating conditions such as temperature, species concentrations and

flow velocity are investigated.The simulation of the system is carried out by the 1D+1D

model for the monolith converter described in section 3.1 and the reactions kinetics as

discussed in section 3.2. In the beginning, the grid independency of the solution is pre-

sented and then the effect of intra-porous diffusion on the system performance and

model prediction is discussed.

ZeoliteAfterward a calibration procedure is proposed to determine the physical and

chemical properties of the system based on experimental data for an Fe-BEA Zeolite-

coated monolith under different conditions as explained in Chapter 4.

The model is verified in two main categories which can be described as follows. The

first category is a set of transient responses of the SCR system at different operating

conditions until the system reaches the steady state again. The second category presents

the response of the system in highly transient conditions representing the automotive

Diesel engine operation.

5.1 Grid independency

In the first step of the numerical simulation, the required resolution of the domain is as-

sessed for the channel and the wall as presented in Figure 5.1 for a step inflow of

500 ppm of NH3. In the upper part, the effect of the resolution along the channel on the

outlet NH3 is investigated. It is started from 10 equidistance longitudinal cells for the

channel of 40.2 mm length and increased by the factor of two. It is shown that changing

of the result for the grids more than 40 is negligible.

Page 69: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

54 Chapter 5. Calibration procedure and results

Figure 5.1. Grid independency analysis, GHSV = 50,000 at 200 ºC and 500 ppm of NH3, calculated

outlet NH3 for different cell numbers (nx) along the channel (upper), calculated NH3 distribution

and surface coverage within the wall for different cell numbers (ny) perpendicular to the wall

(middle), calculated outlet NH3 for different mesh resolution (lower)

0

100

200

300

400

500

0 10 20 30 40 50 60

oule

t NH

3(p

pm)

t (s)

nx80

40

20

10

0.00

0.02

0.04

0.06

0.08

0.10

0.000

0.002

0.004

0.006

0.008

0.010

0.012

0.014

0 5 10 15

NH

3su

rfac

e co

vera

ge [-

]

NH

3*co

ncen

trat

ion

[mol

/m3 ]

t (s)

ny 20 10 5

0

100

200

300

400

500

0 10 20 30 40 50 60

oule

t NH

3(p

pm)

t (s)

nx.ny40.10

40.20

40.30

80.30

Page 70: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 5. Calibration procedure and results 55

The middle part of the Figure 5.1 presents the effect of the resolution, in the direction

perpendicular to the wall, on the calculated ammonia concentration within the pores and

NH3 surface coverage on the catalytic surface in a point on the symmetry axis inside the

wall. It is interpreted that by using 20 equidistance cells in the direction of the wall with

thickness of 82.5 µm, the results are independent of the grids.

The lower part of the Figure 5.1 compares the combinations of different resolutions for

the channel and the wall. Two high resolutions result in almost the same NH3 outlet and

the difference with 40.20 is very small. Since the transient simulation of coupled con-

vective, diffusive and heterogeneously reactive problem is expensive, the resolution of

40 to 20 cells along the channel and inside the wall, respectively, is selected.

5.2 Wall diffusion

The role of intra-porous diffusion in performance of the SCR NOx conversion is as-

sessed by Figure 5.2. Measured NOx conversions from experiments in [29] are com-

pared with predictions of a simplified 1D model, which neglects intra-porous diffusion

and results of a 1D+1D model presented in [29]. Figure 5.2 in the left, shows NOx re-

duction with equal feed of NO and NO2 at different temperatures. The prediction of 1D

model (solid line) is close to the measured data and results from 1D+1D simulation.

On the contrary, in the right part of Figure 5.2 for the cases with lower NO2 feed, the

prediction of the 1D model is higher than measured data especially at low temperatures;

however the 1D+1D simulation has good results. It demonstrates that intra-porous dif-

fusion limits the NOx conversion in some conditions. The difference between 1D and

1D+1D models is explained by the fact that in the 1D simulation without diffusional

limitations, NH3 is absorbed more efficiently within the washcoat, consequently NOx

reacts with adsorbed NH3 more effectively. Therefore a higher NOx conversion is re-

sulted comparing to the experimental data.

Page 71: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

56 Chapter 5. Calibration procedure and results

Figure 5.2. Comparison of NOx conversion calculated in 1D without intra-porous diffusion (solid

lines) with experimental data from [29] (symbols) and 1D+1D simulation considering intra-porous

diffusion [29] (dashed lines), GHSV=36000 h-1 and 500 ppm of NH3 and NOx, left: NO2/NO=1

(green), right: NO2/NO=1/2 (red) and NO2/NO=0/1 (blue)

Figure 5.3 compares the performance of NOx conversion predicted by 1D and 1D+1D

simulations at 300 ºC and similarly shows that the NOx conversion rate predicted by the

1D simulation is higher since the diffusional limit is not considered.

Figure 5.3. Comparison of NO reduction resulted by 1D+1D simulation considering intra-porous

diffusion (solid line) and 1D model without intra-porous diffusion (dashed line)

at 300 ºC for 1000 ppm of NH3 and NO

70

80

90

100

110

150 200 250 300 350 400 450

NO

xco

nver

sion

(%)

T (ºC)

0

20

40

60

80

100

150 200 250 300 350 400

NO

xco

nver

sion

(%)

T (ºC)

0

20

40

60

80

0 10 20 30 40 50 60

NO

con

vers

ion

(%)

t (s)

Page 72: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 5. Calibration procedure and results 57

Figure 5.4. Different regimes controlling the performance of NOx conversion

at different temperatures

Generally, the influence of the intra-porous diffusion on the NOx conversion and the

NH3 slip decreases with increasing the reaction temperature. Figure 5.4 illustrates dif-

ferent regimes in SCR operation and the phenomenon, which control the NOx conver-

sion. At low temperature, called the kinetic regime, the surface reactions are very slow

and the kinetics resistance is clearly dominant and governs the overall rate of NOx con-

version. At high temperature, the external mass transfer resistance becomes prevailing

and the monolith operates under the external mass transfer controlled regime [73]. At

the temperatures in between, diffusional resistance is dominant and the intra-porous dif-

fusion limits the NOx conversion.

It is concluded that the intra-porous diffusion should be included in the SCR reactor

model in order to describe well the effective reaction rates under all conditions varying

from chemical control to external mass transfer control [29].

5.3 Calibration procedure

Ten surface reactions are taken into account for the Fe-Zeolite catalyst as summarised in

Table 3.1. The parameters of the reaction rates, consisting of pre-exponential factors,

activation energies and equilibrium constants, have to be calibrated for each specific

catalyst to satisfy its behaviour and be used for the simulation. In addition to the rate

0

20

40

60

80

100

0 100 200 300 400 500

NO

con

vers

ion

(%)

T (ºC)

Washcoat diffusion

Bulk mass transfer

Kinetics

Page 73: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

58 Chapter 5. Calibration procedure and results

parameters, properties of the catalytic layer, i.e. ammonia adsorption capacity, washcoat

porosity, surface scaling factor and intra-porous effective diffusivity need to be deter-

mined to simulate the response of the system. Table 5.1 lists the parameters of the sys-

tem which are calibrated for the catalyst in use.

Table 5.1. System parameters

(1) Ammonia adsorption 01k eq. (3.15)

(2) Ammonia desorption 02k , E2,, γ eq. (3.16)

(3) Standard SCR

03k′ , E3, 3NHK ,

2OK eq. (3.22)

03k , E3, 2OK eq. (3.24)

(4) Fast SCR 0

4k′ , E4, κ eq. (3.23)

04k , E4 eq. (3.25)

(5) Ammonia spill-over 05k , KLH eq. (3.18)

(6) NO oxidation 06k , E6, a, b eq. (3.29)

(7) Ammonia oxidation 07k , E7 eq. (3.27)

(8) Direct NO2 reduction 08k , E8 eq. (3.31)

(9) Ammonium nitrate formation 09k , E9 eq. (3.34)

(10) Nitrous oxide formation 010k , E10 eq. (3.35)

Ammonia adsorption capacity Ωc eq. (3.20)

Washcoat porosity ε eq. (3.20)

To improve the efficiency of the calibration procedure, the system parameters are cate-

gorized based on their relevance at different conditions, i.e. temperatures and inlet com-

positions. This allows for a step-by-step calibration of separate groups of parameters by

Page 74: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 5. Calibration procedure and results 59

means of the specifically designed experimental data, as explained in Chapter 4. As a

consequence, the numbers of parameters to be calibrated simultaneously is drastically

reduced.

The parameters are determined in an optimization procedure by minimizing the error

between the predicted species concentrations at the outlet and the experimental values.

Since the transient response of the SCR system is important, optimization is carried out

during a step of transient operation. A single error is calculated in each experimental

data point and the total error equals the summation of the absolute single errors for all

species during one step.

The sequence can be outlined as follows: First, parameters relating to ammonia adsorp-

tion and desorption are fitted in the absence of SCR reactions, i.e. by removing NOx

from the inlet flow. As the result of this step, the ammonia adsorption capacity of the

catalyst and the porosity of the washcoated layer are determined as well as the adsorp-

tion/desorption rate parameters.

In a second step, the parameters relating to NO oxidation are calibrated which are inde-

pendent of the other parameters. At least two temperatures are needed to determine the

equilibrium parameters for each NO/NOx ratio in the absence of SCR reactions, i.e. no

ammonia feed.

Once these values are determined, calibration of the parameters for the rest of the reac-

tions can be performed all together. To reduce the numbers of parameters which need to

be simultaneously optimized, small groups of parameters are grouped together at specif-

ic operating conditions, as is explained in detail in the following sections.

5.3.1 Ammonia adsorption/desorption

Parameters of the ammonia adsorption/desorption such as adsorption capacity and po-

rosity of catalyst and rate parameters of adsorption and desorption reactions are cali-

brated based on measured data of the system without NOx in the feed. The parameters

Page 75: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

60 Chapter 5. Calibration procedure and results

are the results of an optimization procedure with the goal of minimizing of the total er-

ror according to equation (5.1) over the measuring points.

3 3

00 3

, ,exp

,exp

steadyn ntNH num NH

tn t NH

X Xerror

X=

−= ∑ (5.1)

Where XNH3,num is predicted and XNH3,exp is derived from the experimental outlet ppm of

ammonia during a step feed till reaching the steady condition. Calibration of the param-

eters is done by a single objective genetic algorithm optimization method [74]. The pro-

cedure is run in a cluster with 64 CPUs by calculating 64 individuals simultaneously

(using one CPU for each individual) enabling rapid optimization for each condition.

Figure 5.5 presents the ammonia adsorption/desorption behavior in the system without

NOx in the feed. Although in this experiment the feed includes Oxygen, according to the

composition of the exhaust gas mixture in the experiments mentioned in Chapter 4,

ammonia oxidation is not observed while the temperature is low. There is a delay in

NH3 concentration at the outlet to reach the inlet concentration, which shows the ad-

sorption of ammonia within the layer (hatched area in Figure 5.5).

Figure 5.5. Step feed of ammonia at the inlet and the system response at the outlet at 200 ºC and

GHSV = 50,000 h-1 for calibration of NH3 adsorption, desorption related parameters.

0

100

200

300

400

500

0 300 600 900 1200 1500 1800

NH

3(p

pm)

t (s)

outlet

inlet

Page 76: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 5. Calibration procedure and results 61

The calibration criteria is equal moles of adsorbed ammonia in both measurement and

simulation as well as a minimum total absolute error of the calculated curve during the

transient operation, for which the parameters are determined in an optimization proce-

dure. The ammonia adsorption capacity of the catalytic washcoat layer is calculated in a

calibration procedure for adsorption/desorption parameters shown schematically in Fig-

ure 5.6.

Figure 5.6. Calibration procedure of ads/des parameters.

The process starts with experimental data and initial guesses for the parameters (square

boxes in Figure 5.6). The value for the effective diffusivity within the washcoat has

been taken from previous studies [43, 53] as a first assumption. Based on the experi-

mental data, the number of adsorbed mole of ammonia is calculated from the colored

area in Figure 5.5 and the inlet conditions like the gaseous flow rate and the tempera-

ture. nsteady denotes the moles of ammonia absorbed on the surface till steady state

reached. By knowing nsteady and the surface coverage of ammonia in steady state, the

Exp. data

Mole of adsorbed Ammonia nsteady

Yes

kads, kdes, Edes, γ

Ωc

Mass of Catalyst θsteady

ε

Transient simulation

Mole of adsorbed Ammonia

Equal No kads, kdes,

Edes, γ, ε, Ωc

Page 77: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

62 Chapter 5. Calibration procedure and results

adsorption capacity is calculated. The steady surface coverage θsteady depends on the re-

action rates of ammonia adsorption/desorption and is calculated based on the assumed

parameters in a transient procedure. The adsorption capacity and the porosity have no

effect on θsteady and just control the rate of the system. The ammonia adsorption capacity

per mass of catalyst, Ωc is found considering the number of moles nsteady, the active

mass of catalyst and the steady surface coverage θsteady. The washcoat porosity can be

adjusted to obtain equal areas (the ammonia moles) in experimental data and numerical

result.

The adsorption capacity is calculated at different temperature and presented in Figure

5.7. It is evident that the ammonia adsorption capacity of the catalyst decreases when

temperature increases, which is in agreement with findings reported in the literature, e.g.

[31].

Figure 5.7. Ammonia adsorption capacity of the catalyst at different temperatures.

The remaining parameters for adsorption/desorption appearing in the reactions rates

(3.15) and (3.16) are summarized in Table 5.2. These values are in the same range as

reported in the literature for similar catalysts [29]. As opposed to the ammonia adsorp-

tion capacity, these are not functions of temperature and remain fixed throughout the

0.0

0.1

0.2

0.3

0.4

0.5

0.6

150 200 250 300 350 400 450 500

Ω(m

mol

/gr c

at)

T (ºC)

Page 78: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 5. Calibration procedure and results 63

remaining investigations. In line with the exothermicity of the adsorption process, the

storage capacity decreased with increasing temperature [31].

Table 5.2. Calibrated adsorption/desorption parameters.

01k = 4.00E+03

02k = 2.00E+09 E2 = 1.20E+05 [J/mol] γ = 0.5

The porosity of the washcoat layer, equals 0.73, is resulted from the calibration proce-

dure of ammonia adsorption/desorption as presented in Figure 5.6.

Figure 5.8 shows the transient response of the system to a step feed of ammonia at

200 ºC and 250 ºC. The numerical simulation of the system response with the calibrated

parameters accurately describes the experimental behaviour. At higher temperature the

outlet ammonia in steady state is lower than inlet ammonia due to oxidation in the pres-

ence of Oxygen. The oxidation of ammonia can be avoided by using inflow feed with-

out Oxygen for ammonia adsorption/desorption study.

Figure 5.8. Step feed of ammonia at the inlet and the system response at the outlet at 200 ºC (left) and 250 ºC (right) and a GHSV = 50,000 h-1, dotted line: outlet from experiment, solid line: outlet

from simulation, dashed line: inlet.

0

100

200

300

400

500

0 200 400 600 800 1000

NH

3(p

pm)

t (s)

0

100

200

300

400

500

0 200 400 600 800 1000

NH

3(p

pm)

t (s)

Page 79: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

64 Chapter 5. Calibration procedure and results

5.3.2 NO oxidation

The reaction rate of NO oxidation, eq. (3.29), has four parameters, which are calibrated

based on the experimental data at the beginning of each test, when there is no ammonia

in the feed. Figure 5.9 shows the measured steady NO2/NOx ratio in fifteen experi-

mental cases for different inlet NO/NOx feed ratios and the catalyst temperature. The

thermodynamic equilibrium between NO and NO2, reaction (3.28), is plotted also as

dashed curve which has been computed based on the Gibbs energy change for the flow

composition at a given temperature (Enthalpy change (∆H) = -58.010 kJ/mol and Entro-

py change (∆S) = -89.429 J/mol/K for 500 ppm NO, 10% O2, 5% H2O and N2 balance).

The equilibrium calculations predict decreasing NO2 fractions for higher temperatures

due to favored NO2 dissociation. It is obvious from Figure 5.9 that the system is far

from equilibrium conditions at steady-state operation, since at temperatures lower than

350 ºC the conversion rises for increasing temperature, limited by kinetics and mass

transfer. At high temperature, however, the conversion is limited by reaction equilibri-

um and is thermodynamically controlled.

Figure 5.9. Steady-state ratio of NO2/NOx measured for three different ratios of NO/NOx in the feed

at five different temperatures, together with thermodynamic equilibrium curve

0.0

0.2

0.4

0.6

0.8

1.0

150 200 250 300 350 400 450 500

NO

2/N

Ox

T (ºC)

NO/NOx=50%

NO/NOx=75%

NO/NOx=100%

Therm. Eq.

Page 80: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 5. Calibration procedure and results 65

The effect of inlet NO concentration was investigated in [75-77], where it was seen that

at the same temperature and NO/NO2 ratio, the conversion efficiency became lower as

the NO concentration increased. This effect was caused by NO self-inhibition and is

commonly observed in the NO oxidation process over precious metal catalysts. Howev-

er this fact was not observed by this study since the temperature and the ratio of

NO/NO2 were different by changing NO concentration.

The semi-equilibrium constant, ( )exp ( ) / ( )eqK a bT RT= − −

, is defined to comprise the

kinetic effects and mass transfer limits of the system. As the steady states varies for dif-

ferent concentrations, the parameters a and b required calibration for each NO/NO2 feed

ratio. Figure 5.10 depicts the linear trend of ln( )eqK

as a function of inverse tempera-

tures (T-1) from which the parameters a and b are derived as the slope and y-intercept,

respectively, according to experimental data. The semi-equilibrium constant

2 2

1 1/2eq NO NO OK C C C− −=

is calculated according to eq. (3.29) from the steady state concen-

trations of NO and NO2 measured in the different ratios of NO/NOx at the different cata-

lyst temperatures.

Figure 5.10. Semi-equilibrium constant calculated based on measured concentrations in steady

state for different NO/NOx ratios at different temperatures (symbols), GHSV = 50,000 h-1.

0.0

1.0

2.0

3.0

4.0

1.0E-03 1.5E-03 2.0E-03 2.5E-03

-ln(K

eq)

1/T (1/K)

NO/NOx=50%

NO/NOx=75%

NO/NOx=100%

Page 81: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

66 Chapter 5. Calibration procedure and results

Table 5.3. Calibrated NO oxidation parameters, GHSV = 50,000 h-1

06k = 15.00 E6 = 2.90E+04 [J/mol]

NO/NOx a b

100 % 2.16E+04 12.99

75 % 3.64E+03 -4.92

50 % 2.75E+03 2.06

After defining the semi-equilibrium condition for each case and determination of the

parameters a and b, the pre-exponential factor and the activation energy of the NO oxi-

dation rate, eq. (3.29), are found based on the measured steady state concentrations of

NO and NO2 in the absence of ammonia. These two parameters are global and valid for

all the conditions. The calibrated parameters for NO oxidation are listed in Table 5.3.

The dependency of the NO oxidation on temperature and the ratio of NO/NOx is well

predicted by the model, as is demonstrated in Figure 5.11.

Figure 5.11. NO oxidation (symbols: experiment, line: simulation) for 500 ppm NO, 10% O2, 5%

H2O and N2 balance at GHSV = 50,000 h-1.

0.0

0.2

0.4

0.6

150 200 250 300 350 400 450 500

NO

2/N

Ox

T (ºC)

NO/NOx=50%

NO/NOx=75%

NO/NOx=100%

Page 82: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 5. Calibration procedure and results 67

5.3.3 Model I and II: Standard and fast SCR reactions

After calibration of ammonia adsorption/desorption and NO oxidation, the rate parame-

ters of the SCR reactions are determined. Two different expressions for the standard

SCR rate proposed by [58, 59] have been assessed. Model II implements the reaction

rate of standard SCR for quasi-equilibrium conditions on the surface, eq. (3.22) as dis-

cussed in section 3.2.2 [58], and in model I the reaction rate is simplified with the same

basis as before, eq. (3.21) [59]. The rate of fast SCR reactions is taken according to eq.

(3.23). The assessment has been carried out at the lowest temperature, 200 ºC, since the

exponential term drops out of rate reaction expressions, eq. (3.37), and hence the influ-

ence of the remaining terms can be highlighted. At low temperatures, the decrease in the

NO concentration after shutting off ammonia due to an inhibition effects is also very

pronounced. This effect is caused by stored NH3 on the catalyst surface that efficiently

reacts with NOx after turning of ammonia and before completely desorbing from the

surface [58].

Figure 5.12. Comparison of measured and predicted response of the system by different Standard

SCR rate expression, model I, blue [59], and model II, red [58]: GHSV = 50,000 h-1, 200 ºC, 500

ppm NOx, 100% NO, symbols: experiment, solid line: simulation, dashed line: inlet.

0

50

100

150

200

250

300

350

400

450

500

0 500 1000 1500 2000

NH

3(p

pm)

NO

(ppm

)

t (s)

NH3,in

NO

NOin

NH3

Page 83: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

68 Chapter 5. Calibration procedure and results

The comparison of the two different rate expressions for the standard SCR reaction is

illustrated in Figure 5.12. It is clearly visible that the non-simplified reaction rate in

model II better predicts the ammonia inhibition effect and the amount of ammonia slip

of the system.

The two reaction rates for standard SCR are compared in Figure 5.13. For equal steady-

state ammonia coverage, the original reaction rate (non-simplified) [58] shows a higher

difference between the maximum rate and the rate at steady-state surface coverage,

which causes a higher inhibition effect.

The parameters of the standard SCR rate expression in model II, eq. (3.22) which has

better prediction for ammonia inhibition effect, is calibrated by the following procedure.

The calibration is based on experimental data of three different ratios of NO/NO2 in the

feed with the same NOx concentration at different temperatures, like shown in Figure

4.3.

Figure 5.13. Standard SCR rate expression by model II in solid line [58], and model I (simplified) in

dashed line [59], at GHSV = 50,000 h-1, 200 ºC, 500 ppm of NOx, 100% NO.

Due to the nature of the rate expressions for the standard and fast SCR reactions, eqs.

(3.22) and (3.23), activation energies are omitted at 200 ºC. Ammonia oxidation is also

insignificant at low temperature [54]. Using a reference temperature for the activation

0.00

0.01

0.02

0.03

0 0.5 1

f(θ

)

θ

Inhibition Effect

Page 84: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 5. Calibration procedure and results 69

energy in the definition of the rate expressions is a practical idea to omit the activation

energy for one temperature and to calibrate the other parameters. The reference temper-

ature is a good starting point for standard and fast SCR. In addition, it is helpful to use

conditions, where one reaction is dominant, such as 100% NO for standard SCR and

NO/NO2 = 1 at low temperature for fast SCR. Consequently, at 200 ºC, the number of

parameters to be optimized drops to five: two pre-exponential factors, KNH3, KO2 and κ.

First optimization is done for the case of 100% NO in the feed to capture the ammonia

inhibition effect. Simulations for conditions of 25% and 50% of NO2 in the feed at

200 ºC can be also done by the calibrated parameters from this stage since they have no

new parameter or reaction.

The high temperature regime was investigated as well with temperatures up to 450 ºC.

Ammonia oxidation is significant at this temperature and its rate parameters, exponen-

tial factor and activation energy, are determined based on experimental data. The activa-

tion energies of the standard and fast SCR reactions are also calculated, while the expo-

nential factors and other parameters in the rates of these two reactions come from a cal-

ibration done at 200 ºC.

Consequently all the SCR reaction related parameters have been determined as shown in

Table 5.4 and are of the same order as in previous studies, cf. e.g. [31]. Results of the

SCR simulation using these calibrated data and corresponding reaction rates, eqs. (3.22)

and (3.23), are presented in section 5.4.1 at low and high temperature.

Table 5.4. Calibrated parameters for standard eq. (3.22), and fast SCR reaction, eq. (3.23)

Standard SCR 03k′ = 9.15E+01 E3 = 6.32E+04

3NHK = 5.50E+02 2OK = 156.39

Fast SCR 04k′ = 8.27E+08 E4 = 5.83E+04 κ = 2.93E+03

Page 85: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

70 Chapter 5. Calibration procedure and results

5.3.4 Model III: Standard and Fast SCR and ammonia spill-over

The most recent expression for the standard and fast SCR reactions, eqs. (3.24) and

(3.25), have been assessed and are compared with the two kinetics of model I and II dis-

cussed before in Figure 5.12. The NOx reduction simulation resulted by different surface

kinetics models and the experimental data are presented in Figure 5.14 at low tempera-

ture of 200 ºC for 500 ppm of NOx (100% NO) and a step of ammonia in the feed, at a

GHSV of 50,000 h-1.

Figure 5.14. Simulation of transient NO reduction with different kinetics of simplified model I

based on [59] (blue), model II based on [58] (green) and dual-site model III based on [63] (red): GHSV = 50,000 h-1, 200 ºC, 500 ppm NOx, 100% NO, symbols: experiment, solid line: simulation,

dashed line: inlet.

The coefficients of determination R2 equal 0.955, 0.966 and 0.974 respectively for mod-

el I based on the kinetics of [59], the blue line, and model II the kinetics from [58], the

green line, and the current model III [63], the red line. R square has been calculated in

the time period presented in Figure 5.14, comprising the step feed of ammonia and the

time needed to reach the steady state after turning off ammonia. R square is calculated

0

50

100

150

200

300

350

400

450

500

4150 4650 5150 5650 6150

NH

3,in

(ppm

)

NO

(ppm

)

t (s)

NH3,in

Exp.

NOin

Page 86: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 5. Calibration procedure and results 71

based on the ratio of the square error, between numerical results and the experimental

data, and the square variation of the measurements. R square represents how effective

the numerical simulation is at capturing the system response and the higher it is the

more effective is the model. It is concluded in Figure 5.14 that the new kinetics of the

third model considering a dual-site mechanism significantly improves the predictions of

steady state and in particular the response of the system when ammonia is shut off and

NO reduction temporarily increases due to the absorbed ammonia inhibition effect.

After testing the different kinetics for NOx reduction by ammonia SCR and comparing

the results, the best and most recent kinetics are selected and employed to model the

NH3-SCR system. Model III successfully simulates the transient response of the system

with the good predictions of steady state and ammonia inhibition effect. In the selected

model, model III, standard and fast SCR reactions are considered according to eq. (3.24)

and eq. (3.25) respectively, accompanied by rate of NH3 spill-over as eq. (3.18).

The model parameters are determined in an optimization procedure by minimizing the

error between the predicted species concentrations at the catalyst outlet and the experi-

mental values. Since the transient response of the SCR system is important, optimiza-

tion is carried out for the transient test conditions. A single error is calculated for each

experimental data point and the total error equals the summation of the absolute single

errors for all species during one step. The duration of transient periods varies for differ-

ent test conditions, e.g. at high temperatures or high NO2 fractions in the feed, the sys-

tem reacts very fast and rapidly reaches the steady state after a very short transition

time. Therefore, the measurement of steady state concentrations is sufficient in this case

for a successful calibration. However, for low temperatures the entire process is slow

and system transients have to be considered due to changes in the ammonia feed con-

centration and the ammonia inhibition effect. The transients need to be simulated and

accounted for in error calculations.

The activation energies of the reactions are omitted at 200 ºC, as can be seen from their

respective rate expressions given in Table 3.1. Furthermore, at this low temperature

Page 87: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

72 Chapter 5. Calibration procedure and results

ammonia oxidation and direct NO2 reduction are also insignificant and neglected as

proposed in [54]. Therefore, the number of parameters at 200 ºC is minimal and this

temperature is selected as the starting point for the calibration procedure. Standard SCR

is the main reaction when the NOx in the feed is only NO and for NOx = 50% NO and

50% NO2, NOx is mainly reduced by Fast SCR. Both SCR reactions should be calibrat-

ed such that all conditions in between Standard and Fast SCR are also captured.

2OK in the Standard SCR rate, eq. (3.24), is the ratio of the rate constants of DeNOx

reaction between redox and acidic sites and re-oxidation of the redox site. Increasing of

this parameter magnifies the inhibition effect which is modelled by the ammonia spill-

over reaction. KLH in the rate of ammonia spill-over, eq. (3.18), represents the equilib-

rium constant of the reversible ammonia spill-over step. This parameter regulates the

rate of NOx reduction and the inhibition effect. The pre-exponential factor in the rate of

this reaction, 05k in eq. (3.18), determines the inhibition effect but does not change the

NOx reduction.

Due to the definition of the reaction rate coefficients by eq. (3.37), the pre-exponential

factors ( 0ik ) are already determined at 473 K for the next step with no effect on the acti-

vation energies. However, the activation energies have to be calibrated for higher tem-

peratures.

The process of NOx reduction is mainly determined by the Standard and Fast SCR reac-

tion accompanied by ammonia spill-over. Standard SCR is very active at high tempera-

ture around 450 ºC and can be calibrated independent of Fast SCR and ammonia inhibi-

tion under these conditions when there is only NO in the feed. At low temperature of

250 ºC, both Standard and Fast SCR participates in NOx reduction and Fast SCR is

more effective with higher ratio of NO2/NOx in the feed. The calibrated rate parameters

are reported in Table 5.5.

Page 88: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 5. Calibration procedure and results 73

Table 5.5. Rate parameters of standard and fast SCR and NH3 spill-over

Standard SCR 03k = 5.50E+02 E3 = 5.00E+04

2OK = 5.50E+02 eq. (3.24)

Fast SCR 04k = 1.00E+06 E4 = 8.00E+04 - eq. (3.25)

NH3 spill-over 05k = 5.00E-01 - KLH = 3.90E+00 eq. (3.18)

5.3.5 NH3 oxidation and reactions between NO2 and NH3

Ammonia oxidation, reaction (3.26), and direct NO2 reduction, reaction (3.30), showed

only a small activity at low temperatures and are therefore calibrated at high tempera-

ture. The calibration is carried out for the condition of 100% NO in the feed, where

formation of ammonium nitrate and nitrous oxide are not active (as discussed below)

and their parameters do not participate in the simulation.

Table 5.6. Calibrated rate parameters

Ammonia oxidation 07k = 5.00E-02 E7 = 3.06E+04 eq. (3.27)

Direct NO2 reduction 08k = 3.00E+02 E8 = 4.00E+04 eq. (3.31)

Ammonium nitrate formation 09k = 1.00E+04 E9 = 1.50E+04 eq. (3.34)

Nitrous oxide formation 010k = 1.00E+02 E10 = 2.00E+04 eq. (3.35)

The formation of ammonium nitrate, reaction (3.32), and nitrous oxide formation, reac-

tion (3.33), started to become significant for NO2,in/NOx,in ≥ 50% in the feed. The re-

lated model parameters are determined for NO2,in/NOx,in = 50% at a low temperature,

where ammonia oxidation and NO2 direct reduction could be neglected. Ammonia oxi-

dation should be calibrated such that 10 ppm ammonia slip out of the system is allowed

especially for high dosage of ammonia. The calibrated reaction rate parameters are

listed in Table 5.6.

Page 89: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

74 Chapter 5. Calibration procedure and results

5.4 Transient operation and steady state

The models developed in Chapter 3 and calibrated parameters from section 5.3 are eval-

uated by comparing the simulation results and measured data. The first set of experi-

ments present transient response of the SCR system until gaining steady state, as men-

tioned before in Chapter 4.

5.4.1 Model II

Comparison of the two first models in section 5.3.3, indicates that the simulation using

model I is not successful in capturing the ammonia inhibition effect. Therefore the study

is carried out with model II due to the significantly better predictions of the ammonia

inhibition effect. The simulation results obtained by model II are presented at low and

high temperatures and the response of the system is discussed under different condi-

tions.

Figure 5.15 in the upper part shows the numerical and experimental results for the outlet

concentrations at 200 ºC without NO2 in feed. Outlet NO in the absence of NH3 is lower

than the feed because of the oxidation of part of the NO to NO2 in the system, which is

predicted well by the model. Since the NO2 concentration is low under this condition,

the SCR reaction is dominated by the standard SCR. The response of the system is well

predicted when compared to experimental data. The middle part of Figure 5.15 shows

the outlet concentration when 75% of the NOx in the feed is NO. The fast SCR reaction

is more effective in this case and ammonia inhibition is not observed. There is a good

agreement between the experiment and the simulation, which confirms validity of pa-

rameters for different inlet concentrations. Figure 5.15 in the lower part presents the re-

sponse of the system with equal amounts of NO and NO2 in the feed. Here fast SCR is

the dominant reaction, while standard SCR is almost negligible and there is no effect of

ammonia inhibition. At low temperatures NOx reduction reaches an acceptable level by

help of the fast SCR reaction due to the presence of NO2 in feed. Good agreement was

reached between numerical simulation and experiment in this case as well.

Page 90: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 5. Calibration procedure and results 75

Figure 5.15. Transient operation at GHSV = 50,000 h-1, 450 ºC, 500 ppm NOx, NO/NOx = 100%

(upper), 75% (middle), 50% (lower), NH3/NOx = (0.8, 1.0, 1.2) times of the optimum dosage (αOD for

10 ppm ammonia slip), Symbols: experiment, line: simulation using model II, dashed line: inlet,

red: NO, blue: NO2, green: NH3.

0

50

100

150

200

250

300

200

250

300

350

400

450

500

0 2000 4000 6000 8000

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

NOin

NH3,in

NOout

NO2,out

NH3,out

0.8αOD

1.0αOD

1.2αOD

0

100

200

300

400

500

0

50

100

150

200

250

300

350

400

750 1750 2750 3750 4750 5750

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

NOin

NH3,in

NOout

NO2,out

NH3,out

NO2,in

0.8αOD

1.0αOD

1.2αOD

0

100

200

300

400

500

600

700

800

0

50

100

150

200

250

300

6200 6700 7200 7700 8200 8700 9200

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

NOin

NH3,in

NOout

NO2,out

NH3,out

NO2,in

0.8αOD

1.0αOD

1.2αOD

Page 91: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

76 Chapter 5. Calibration procedure and results

Figure 5.16. Transient operation at GHSV = 50,000 h-1, 450 ºC, 500 ppm NOx, NO/NOx = 100%

(upper), 75% (middle), 50% (lower), NH3/NOx = (0.8, 1.0, 1.2) times of the optimum dosage (αOD for

10 ppm ammonia slip), Symbols: experiment, line: simulation using model II, dashed line: inlet,

red: NO, blue: NO2, green: NH3.

0

100

200

300

400

500

600

0

100

200

300

400

500

50 250 450 650 850 1050

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

NOin

NH3,in

NOout

NO2,out

NH3,out

0.8αOD

1.0αOD

0

100

200

300

400

500

600

700

0

50

100

150

200

250

300

350

400

100 300 500 700 900 1100 1300

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

NOin

NH3,inNOout

NO2,out

NH3,out

NO2,in

0.8αOD

1.0αOD

1.2αOD

0

100

200

300

400

500

600

700

800

0

50

100

150

200

250

300

1900 2100 2300 2500 2700 2900 3100 3300

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

NOin

NH3,in

NO2,out

NOout

NH3,out

NO2,in

0.8αOD

1.0αOD

1.2αOD

Page 92: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 5. Calibration procedure and results 77

Figure 5.16 in the upper part shows the response of the system at high temperatures.

Although there is no NO2 in the feed, the standard SCR reaction is fast enough at this

high temperature to reduce NO almost completely. Ammonia consumption and NOx re-

duction is much higher compared to the low temperature range. There is also no ammo-

nia inhibition effect observable at high temperatures. Again, good agreement can be

seen between numerical results and experimental data. To verify these parameters, two

further variations of the inlet concentration are also simulated. The simulation of the

system at high temperature with 25% NO2 in the feed are presented in the middle part of

Figure 5.16. NO oxidation is lower in this case, because of the equilibrium behavior of

the reaction. Figure 5.16 in the lower part shows the case with equal amounts of NO and

NO2 in feed at high temperature. In this condition, the NO/NO2 equilibrium lies on the

side of NO and part of the inlet NO2 is converted to NO.

During the course of the calibration procedure and the system simulation, model III was

proposed by the literature as presented in 5.3.4. It is concluded from the discussion in

section 5.3.4 that the model III results in better predictions for the response of the sys-

tem, therefore the study is continued by model III and the different operating conditions

are simulated in the following section.

5.4.2 Model III

Comparison of different kinetics in section 5.3.4 concludes that the model III has better

prediction of the SCR system during the transition and in steady state as well as for

ammonia inhibition effect. Consequently the dual-site mechanism, model III, is selected

to simulate the SCR system and the further discussions are presented for simulations

with model III.

The validation of the model is presented in 90 different operating conditions for slow

transient responses of the system. The transient evolution of the conversion following

step feeds of ammonia is studied for three NO/NOx ratios of 100%, 75% and 50% up to

steady state. Ammonia is fed stepwise at three different dosage ratios of α = 0.8, 1.0,

Page 93: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

78 Chapter 5. Calibration procedure and results

1.2 with respect to the optimum dosage αOD, resulting in 10 ppm ammonia slip, as

shown in Figure 4.3.

Figure 5.17 (left) illustrates the NOx reduction efficiency of the system at low space ve-

locity (GHSV = 30,000 h-1) for different NO/NOx ratios and different ammonia inlet

concentrations. The numerical results at steady state are compared to experimental data

and indicated that the model is accurate concerning the steady state NOx reduction of

the system at different temperatures and feed ratios. At low temperature, the presence of

NO2 significantly improves the NOx reduction efficiency due to the Fast SCR reaction.

However, at high temperature Standard SCR is active enough to reduce all the NOx in

the system and the effect of Fast SCR is negligible. Increasing the inlet ammonia con-

centration to reach the optimum dosage ratio enhances the NOx reduction efficiency

while feeding more ammonia than optimum dosage does not change the NOx conversion

in the system. The NOx reduction efficiency slightly decreases at high temperature

(450 ºC). This can happen at high temperature condition because of shorter residence

time and the fact that the rate of ammonia desorption is higher than that of ammonia ad-

sorption; therefore, the surface coverage of ammonia decreases and the NOx reduction

drops.

The operation of the system at the higher space velocity (GHSV = 50,000 h-1) is also

examined and the results are presented in Figure 5.17 (right). NOx reduction is correctly

calculated by the model not only at low but also at high space velocity. Due to the

shorter residence time at higher space velocities, NOx conversion is lower than for slow

gas flow. The influence of the space velocity on NOx reduction is most obvious for

suboptimal reaction conditions, such as low temperatures without NO2 in the feed.

When there is enough NO2 in the system to convert all NOx by the Fast SCR reaction or

at high temperature, NOx conversion is fast and not affected by the GHSV.

An inhibition effect on the SCR performance is observed over the Fe-Zeolite catalyst in

the presence of ammonia. This effect is more obvious at low temperatures and without

NO2 in the feed when the major reaction is Standard SCR. Figure 5.18 exemplarily

Page 94: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 5. Calibration procedure and results 79

shows the transient operation of the system at low temperature of 200 ºC with 100% NO

in the feed. The simulation predicts well the transient response of the system including

the ammonia inhibition effect as well as the steady state NOx conversion of the system.

Figure 5.17. NOx reduction at different temperatures as well as different NO/NOx and NH3/NOx = 0.8, 1.0, 1.2 times of the optimum dosage (αOD for 10 ppm ammonia slip), Left: GHSV=30,000 h-1,

Right: GHSV=50,000 h-1.

Page 95: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

80 Chapter 5. Calibration procedure and results

Figure 5.18. Transient operation at GHSV = 50,000 h-1, 200 ºC, 500 ppm NOx, NO/NOx = 100%, NH3/NOx = (0.8, 1.0, 1.2) times of the optimum dosage (αOD for 10 ppm ammonia slip), Symbols: experiment, lines: simulation with model III, dashed line: inlet, red: NO, blue: NO2, green: NH3.

Figure 5.19. Transient operation at GHSV = 30,000 h-1, 200 ºC, 500 ppm NOx, NO/NOx = 100%, NH3/NOx = (0.8, 1.0, 1.2) times of the optimum dosage (αOD for 10 ppm ammonia slip), Symbols: experiment, line: simulation with model III, dashed line: inlet, red: NO, blue: NO2, green: NH3.

0

50

100

150

200

250

300

200

250

300

350

400

450

500

2000 3000 4000 5000 6000 7000 8000

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

NOin

NH3,in

NOout

NO2,out

NH3,out

0

50

100

150

200

250

300

200

250

300

350

400

450

500

2800 3800 4800 5800 6800 7800 8800 9800

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

NOin

NH3,in

NOout

NO2,out

NH3,out

Page 96: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 5. Calibration procedure and results 81

The inhibition effect is stronger at low space velocity. A longer residence time allows

more NOx conversion by ammonia desorbing from the catalytic layer after shutting off

ammonia, as presented in Figure 5.19 for the transient simulation of the system at

GHSV = 30,000 h-1.

Since the transient operation of the SCR system is most important for automotive appli-

cations, the model is intensively checked for these conditions. Figure 5.20 shows the

NOx reduction at high temperature (450 ºC) with 100% NO in the feed. The Standard

SCR reaction is fast enough to remove NOx from the feed and there is no inhibition ef-

fect at this high temperature.

Figure 5.20. Transient operation at GHSV = 50,000 h-1, 450 ºC, 500 ppm NOx, NO/NOx = 100%, NH3/NOx = (0.8, 1.0, 1.2) times of the optimum dosage (αOD for 10 ppm ammonia slip), Symbols: experiment, line: simulation with model III, dashed line: inlet, red: NO, blue: NO2, green: NH3.

0

100

200

300

400

500

600

0

100

200

300

400

500

350 550 750 950 1150 1350 1550

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

NOin

NH3,in NOout

NO2,out NH3,out

Page 97: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

82 Chapter 5. Calibration procedure and results

Figure 5.21. Transient operation at GHSV = 50,000 h-1, 250 ºC, 500 ppm NOx, NO/NOx = 100%

(upper), 75% (middle), 50% (lower), NH3/NOx = (0.8, 1.0, 1.2) times of the optimum dosage (αOD for 10 ppm ammonia slip), Symbols: experiment with model III, line: simulation, dashed line: inlet,

red: NO, blue: NO2, green: NH3.

0

50

100

150

200

250

300

350

400

100

150

200

250

300

350

400

450

500

1000 1500 2000 2500 3000 3500 4000 4500

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

NOin

NH3,in

NOoutNO2,outNH3,out

0

100

200

300

400

500

0

50

100

150

200

250

300

350

400

1800 2300 2800 3300 3800 4300 4800

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

NOin

NH3,in

NOoutNO2,outNH3,out

NO2,in

0

100

200

300

400

500

600

0

50

100

150

200

250

5600 6100 6600 7100 7600 8100

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

NOin

NH3,in

NOout

NO2,outNH3,out

NO2,in

Page 98: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 5. Calibration procedure and results 83

Figure 5.21 illustrates the influence of the NO/NOx ratio including Fast SCR conditions

at low temperature (250 ºC). This temperature is not high enough to reduce NOx only

with Standard SCR, observable from the only 50% NOx conversion for the case with

100% NO in the feed. Ammonia inhibition is also observed under this condition. Feed-

ing of NO2 enables the Fast SCR reaction and increases the NOx conversion level to

100% for equal shares of NO and NO2 in the feed.

To check the accuracy of the model for different SCR operating conditions, the cumula-

tive NOx mass is calculated for each condition during transient operation and compared

to measured data. Figure 5.22 compares the results for all 30 different cases investigat-

ed, i.e. two different GHSV, three NO/NOx ratios and five temperatures. Each case in-

cludes three dosages of ammonia, (0.8, 1.0, 1.2) times αOD, during the transient opera-

tion. The coefficient of determination R2 is 0.997 and confirms the very good accuracy

of simulation. The results for the simulation of the all 90 transient cases are presented in

the appendix.

Figure 5.22. Cumulative NOx (g) after reaching the steady state at 200, 250, 300, 350, 450 ºC with

NO/NOx = 100, 75, 50 % at GHSV = 50,000 and 30,000 h-1.

0.00

0.05

0.10

0.15

0.20

0.25

0.00 0.05 0.10 0.15 0.20 0.25

num

eric

al

experimental

R2 = 0.997

Page 99: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

84 Chapter 5. Calibration procedure and results

Figure 5.23. Comparison between predicted and experimental ammonia slip at different tempera-tures and different NO/NOx ratios and NH3/NOx = 0.8, 1.0, 1.2 times of the optimum dosage (αOD

for 10 ppm ammonia slip), Left: GHSV=30,000 h-1, right: GHSV=50,000 h-1.

Ammonia slip is a critical issue for SCR systems and is primarily controlled by the cor-

rect dosage of the ammonia precursor urea upstream of the converter, in dependency of

NOx,in, the gas flow rate, catalyst temperature and stored ammonia.

Page 100: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 5. Calibration procedure and results 85

Figure 5.23 compares the predicted ammonia concentration at the catalytic converter

outlet with the measured data. Good agreement can be observed for the different tem-

peratures, both GHSVs as well as NO/NOx and NH3/NOx ratios. It is obvious that am-

monia slip is inversely correlated with the NOx reduction efficiency, see Figure 5.17.

When NOx reduction increases, at high temperature or low NO/NOx ratio more ammo-

nia is consumed and NH3-slip decreases consequently. The slip at optimum ammonia

dosage is similar to that at lower dosage rates, since ammonia is almost totally con-

sumed under these conditions. However, overdosage (1.2 αOD) of ammonia results in

more slip, since NOx reduction cannot be improved significantly by simply dosing more

ammonia (see Figure 5.17). This tradeoff between ammonia slip and NOx reduction in

SCR systems is illustrated by Figure 5.24.

Figure 5.24. Tradeoff between NOx reduction (solid line) and ammonia slip (dashed line)

at GHSV = 50,000 h-1 and NO/NOx = 75%, NH3/NOx = 0.8 (circle), 1.0 (triangle), 1.2 (star) times of the optimum dosage (αOD for 10 ppm ammonia slip)

In summary it can be stated that by comparison of the results of the numerical simula-

tion with the experimental data, the calibrated parameters and the simulation method

can be checked. It proves that the physical properties of the system and reactions rates

parameters are valid for the range of the operating conditions of the SCR system.

0

5

10

15

20

25

30

45

60

75

90

105

150 250 350 450

NH

3sl

ip [%

]

DeN

Ox

[%]

T (ºC)

Page 101: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

86 Chapter 5. Calibration procedure and results

5.5 Highly transient performance

While excellent steady conversion is predicted over a broad range of temperatures,

NO/NOx ratios and velocities, it is shown that ammonia inhibition has a strong effect on

the dynamic behaviour of the system at low temperature. Here, the developed model,

has hence been assessed specifically with respect to its predictive performance at highly

transient conditions representative of automotive Diesel engine operation. Due to the

strongly transient nature of the tests, only the most recent chemical kinetics (i.e.

model III) have been employed for these simulations.

The modelled dynamic response of the system at 200 ºC is presented in Figure 5.25.

NOx reduction was highly transient at this temperature especially for 100% NO in the

feed (upper), however the conversion increased by adding 25% NO2 to the system

(lower). The presence of NO2 and consequently running the Fast SCR speed up the De-

NOx process and the outlet concentrations were approaching steady conversion as ob-

served for slow transient operation, in particular for the longer ammonia pulses, how-

ever with considerable delay.

At 350º C as shown in Figure 5.26, the reactions were much faster and the outlet species

reached steady concentration in the end of the ammonia step feeds even for 100% NO in

the feed (upper). The presence of NO2 and hence Fast SCR further improves the DeNOx

performance as can be seen in the lower half of Figure 5.26.

The dynamic performance of the system was also investigated at the lower space veloc-

ity of GHSV = 30,000. Figure 5.27 compares the predicted and measured response of

the low speed system at high temperature of 450 ºC for two different NO/NOx ratios of

75% and 50%. Due to the longer residence time, NOx reduction was more efficient and

almost steady during each step feed of ammonia. Feeding of NO2 did not change the

performance significantly, since the temperature was high enough and Standard SCR

was sufficient to reduce NOx effectively.

Page 102: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 5. Calibration procedure and results 87

Figure 5.25. Comparison of predicted and experimental system response at dynamic SCR opera-tion: GHSV = 50,000 h-1, 200 ºC, NO/NOx = 100% (upper) and 75% (lower), for three ammonia dosage ratios α = 0.8, 1.0, 1.2 (with respect to the optimum dosage αOD, resulting in 10 ppm slip). Symbols: experiment, lines: simulation using model III, dashed line: inlet, red: NO, blue: NO2,

green: NH3.

0

50

100

150

200

250

300

200

250

300

350

400

450

500

0 1000 2000 3000 4000 5000 6000

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

NOin

NH3,in

NOout

NO2,outNH3,out

0

100

200

300

400

500

600

700

800

900

0

50

100

150

200

250

300

350

400

6900 7900 8900 9900 10900 11900 12900

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

NOin

NH3,inNOout

NO2,out

NH3,out

NO2,in

Page 103: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

88 Chapter 5. Calibration procedure and results

Figure 5.26. Comparison of predicted and experimental system response at dynamic SCR opera-tion: GHSV = 50,000 h-1, 350 ºC, NO/NOx = 100% (upper) and 75% (lower), for three ammonia dosage ratios α = 0.8, 1.0, 1.2 (with respect to the optimum dosage αOD, resulting in 10 ppm slip). Symbols: experiment, lines: simulation using model III, dashed line: inlet, red: NO, blue: NO2,

green: NH3.

0

100

200

300

400

500

600

700

800

0

100

200

300

400

500

0 1000 2000 3000 4000 5000 6000

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

NOin

NH3,in

NOout

NO2,outNH3,out

0

100

200

300

400

500

600

700

800

0

50

100

150

200

250

300

350

400

6900 7900 8900 9900 10900 11900 12900

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

NOin

NH3,in

NOout

NO2,out

NH3,out

NO2,in

Page 104: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 5. Calibration procedure and results 89

Figure 5.27. Comparison of predicted and experimental system response at dynamic SCR opera-

tion: GHSV = 30,000 h-1, 450 ºC, NO/NOx = 75% (upper) and 50% (lower), for three ammonia dos-age ratios α = 0.8, 1.0, 1.2 (with respect to the optimum dosage αOD, resulting in 10 ppm slip). Sym-bols: experiment, lines: simulation using model III, dashed line: inlet, red: NO, blue: NO2, green:

NH3.

0

100

200

300

400

500

600

0

50

100

150

200

250

300

350

400

6900 7900 8900 9900 10900 11900 12900

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

NOin

NH3,inNOout

NO2,out

NH3,out

NO2,in

0

100

200

300

400

500

600

0

50

100

150

200

250

300

350

13500 14500 15500 16500 17500 18500 19500

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

NOin

NH3,inNOout

NO2,out

NH3,out

NO2,in

Page 105: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

90 Chapter 5. Calibration procedure and results

Figure 5.28. Predicted and experimental evolution of cumulated NOx during 20,000 s of fast transi-

ent SCR operation at five different temperatures and three ratios of NO/NOx (50%, 75% and

100%) for a GHSV = 50,000 h-1.

To investigate the performance of NOx reduction under dynamic operation and verify

the simulation results, the cumulative mass of the outlet NOx was calculated for all five

temperatures over 20,000 s of dynamic operation with different ratios of NOin/NOx,in

and varying dosage of ammonia. The NOx reduction in the system is illustrated as the

difference between the cumulative NOx at the inlet and outlet of the system in Figure

5.28. Increasing the temperature enhanced the DeNOx performance, which was highest

at 350 ºC. A further increase in temperature, however, led to a reduction in the DeNOx

performance which is generally observed in SCR systems.

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

0 5000 10000 15000 20000

Cum

ulat

ive

unco

nver

ted

NO

x(g

)

t (s)

inlet200 ºC,exp200 ºC,num250 ºC,exp250 ºC,num300 ºC,exp300 ºC,num350 ºC,exp350 ºC,num450 ºC,exp450 ºC,num

NO/NOx = 100%

NO/NOx = 75%

NO/NOx = 50%

Page 106: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 5. Calibration procedure and results 91

The reason for this behaviour could be ammonia dosage and gas phase diffusion limita-

tion due to too short residence times in the catalytic converter as well as less ammonia

storage at high temperature. Since the GHSV is defined at standard conditions, increas-

ing temperatures resulted in faster inlet velocity due to lower gas densities and conse-

quently significantly lower residence times (defined here as the channel length divided

by the inlet velocity) in the channels of the monolith.

To assess the influences of the different parameters on the conversion at high tempera-

tures, five different test cases at 350 and 450 ºC were studied numerically and summa-

rised in Table 5.7. A single ammonia pulse was studied at 350 and 450 ºC, with a ratio

of NO/NOx of 75 % as shown in Figure 5.29; for the sake of visibility, NO and NO2 are

presented here jointly as NOx and a scale split has been introduced for the region of in-

terest.

Table 5.7. Overview of test cases employed to identify temperature/residence time/ammonia dosage dependence on the NO conversion rate

inlet (ppm) outlet

No. T (ºC) NO NO2 NH3 NOx

1 350 GHSV 50,000 375 125 416, α=0.8αOD 88

2 450 GHSV 50,000 375 125 408, α=0.8αOD 94.2

3 450 GHSV 50,000 375 125 416 87 4 450 equal residence time to No.1 375 125 416 86

5 450 equal residence time to No.1 375 125 408, α=0.8αOD 93.9

Case 1 and case 2 correspond to the pulses starting at ~7,900 s in Figure 5.26 and Figure

5.27 at 350 ºC and 450 ºC, respectively. The increased conversion rate at the higher

temperature can clearly be observed during the transient phase; steady state was attained

much more rapidly for the higher temperature. Once steady state was reached, case 1 at

the lower temperature showed increased conversion, however. Case 3, 4 and 5 were

Page 107: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

92 Chapter 5. Calibration procedure and results

‘synthetic’ test cases where residence time and ammonia dosage were varied independ-

ently with respect to the ‘reference’ case 2.

Figure 5.29. System response to a single pulse of ammonia at 500 ppm NO (NO/NOx ratio 75%) for

the five test cases (denoted 1 to 5) from Table 5.7 to assess the influence of temperature, ammonia

dosage and residence time on the steady NO conversion. Outlet ppm NOx values are for joint NO

and NO2 and use a split scale for clarity.

Comparing case 3 to case 2 shows the dominant influence of inlet ammonia under over-

all ammonia-deficient conditions (α around 0.8) while the flow conditions such as inlet

NOx and residence time were the same: an increase of the ammonia dosage from

408 ppm to 416 ppm resulted in a significant higher NOx reduction. Compared to the

ammonia inlet concentration other parameter variations showed much less influence. It

is interesting to note that an increase of the temperature from 350 ºC to 450 ºC at other-

wise comparable conditions (case 1 and case 3) resulted in almost the same NOx outlet

(88 ppm and 87 ppm), proving the high selectivity of the Fe-Zeolite catalyst even at

high temperatures. The DeNOx performance was also still very similar when the inlet

gas velocity was reduced from case 3 to case 4 with longer residence time equal to case

1. The slight improvement in NOx outlet from 87 ppm to 86 ppm is caused by the better

gas phase diffusion for the longer residence time in case 4. Case 5 assesses the influence

400

500

1 2 3 4 5

80

90

100

110

120

0 50 100 150

outle

t NO

x(p

pm)

T (s)

Page 108: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 5. Calibration procedure and results 93

of the residence time at 450 ºC for the ammonia dosage of 408 ppm. It is evident, that

increasing the residence time to the same duration as for the lower temperature (case 1),

shows again only a small impact on the level of NO conversion. When combining the

ammonia concentration and the residence time of case 1 with the high temperature, the

highest conversion is reached as case 4 clearly demonstrates. It can hence be concluded,

that the influence on NOx conversion is predominantly controlled by the NH3 feed under

these ammonia-deficient conditions and only weakly dependent on the residence time.

Figure 5.30. Predicted and experimental evolution of cumulated NOx during 20,000 s of fast transi-ent SCR operation at five different temperatures and three ratios of NO/NOx (50%, 75% and

100%) for a GHSV = 30,000 h-1.

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

0 5000 10000 15000 20000

Cum

ulat

ive

unco

nver

ted

NO

x(g

)

t (s)

inlet

200 ºC,exp

200 ºC,num

250 ºC,exp

250 ºC,num

300 ºC,exp

300 ºC,num

350 ºC,exp

350 ºC,num

450 ºC,exp

450 ºC,num

NO/NOx = 100%

NO/NOx = 75%

NO/NOx = 50%

Page 109: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

94 Chapter 5. Calibration procedure and results

Figure 5.30 illustrates the accuracy of the simulation by comparing the predicted cumu-

lated NOx with the corresponding experimental results for low space velocity of

GHSV = 30,000 during 20,000 s. The cumulative NOx calculated by the model were in

good agreement with measured data for this velocity as well and the final error was less

than 2 per cent in both cases.

Figure 5.31 (left) presents the outlet cumulative NOx versus temperatures after 20,000 s

of dynamic run for all 90 operating conditions (two space velocities, three NO2,in/NOx,in

ratios, three dosages of ammonia and five temperatures). Excellent predictions can be

seen for both space velocities at all temperatures providing strong evidence for the va-

lidity of the developed model and the chemistry utilised as well as the parameterisation.

NOx reduction was more efficient at low space velocity due to longer residence time,

while the trends of the system response were similar at both space velocities.

Figure 5.31. Cumulative NOx (left) and ammonia slips (right) after 20,000 s of fast transient opera-

tion for different ratios of NO/NOx at different temperatures and space velocities.

Ammonia slip is a critical issue in SCR systems, which requires careful control of the

amount of urea, which is injected upstream of the converter as the source of ammonia.

Figure 5.31 (right) depicts the cumulative ammonia slip after 20,000 s of dynamic op-

eration for all 90 different conditions, as listed above, as a function of temperature. The

30000GHSV,num 30000GHSV,exp

50000GHSV,num 50000GHSV,exp

0.6

0.8

1.0

1.2

1.4

150 250 350 450cum

ulat

ive u

ncon

vert

ed N

Ox

(g)

T (ºC)

0

1

2

3

4

5

0.000

0.005

0.010

0.015

0.020

150 250 350 450

cum

ulat

ive N

H3

slip

(%)

cum

ulat

ive N

H3

slip

(g)

T (ºC)

Page 110: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 5. Calibration procedure and results 95

left axis shows the cumulative mass, while the right axis presents the ratio of cumulative

ammonia slip and total inlet ammonia. The predicted ammonia slip agreed well with the

experimental data. It is also clearly visible that the ammonia slip during dynamic opera-

tion was almost zero at low temperature, which can be explained by the transient behav-

iour of ammonia adsorption/desorption. When the operating conditions were slow

enough to reach the steady state at low temperatures (the 200 ºC case in Figure 5.18) the

ammonia surface coverage reached steady state as well and the slip of ammonia became

equal to the amounts defined by 0.8, 1.0 or 1.2 times αOD. On the contrary, when the

dynamic operation was too fast to reach a steady state for the ammonia surface coverage

especially at low temperatures, the ammonia slip was always lower than the experimen-

tally observed value (Figure 5.25). Since the reactions were much faster at high temper-

ature, the system could reach the steady state even in dynamic operation and therefore

ammonia slip was almost as much as expected (Figure 5.26).

5.6 Non-isothermal operation

Non-isothermal simulation is necessary to investigate the performance of an SCR sys-

tem, since the system mainly operates in transient non-isothermal conditions. The tran-

sient response of the system to the temperature steps in the feed is predicted by a heat

transfer model, as explained in section 3.1, coupled to the reaction kinetics.

To represent the conditions of the experiment, an external heat source is considered at

the wall proportional to the difference between the wall temperature and temperature of

the coil. The external heat source over the peripheral area of the cell is described in eq.

(5.2). Temperature of the wall, T

, is a functional of longitudinal position while the

heating coil temperature, Tcoil, is considered uniform. Temperature of the coil is consid-

ered equal to the inlet temperature which in the beginning is the same as the wall tem-

perature. The second term is hence introduced to provide an additional heating source in

the beginning, which is assumed proportional to difference between the temperature of

the wall and the maximum temperature of the system. The two coefficients in the ex-

Page 111: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

96 Chapter 5. Calibration procedure and results

pression, c1 and c2, are calibrated by the numerical simulation to satisfy the temperature

response of the system in the outlet.

, 1 2 max( ) ( )H x coil x xq c T T c T T= − + −

(5.2)

The inlet temperature and concentrations are presented in Figure 5.32 and the numerical

results are compared with measurements at the outlet. Inlet temperature is ramped from

200 to 450 ºC and inlet ammonia is increased corresponding to the optimum dosage of

ammonia for 10 ppm slip which is the same as the dosage for the system in steady state,

presented in section 5.4.2.

Figure 5.32. Comparison of predicted and experimental system response to the inlet temperature step at GHSV = 50,000 h-1, NO/NOx = 100% and optimum ammonia dosage (resulting in 10 ppm

slip). Symbols: experiment, lines: simulation, dashed line: inlet, red: temperature, blue: NO, green: NH3.

The temperature is predicted correctly by applying the calibrated heat source (c1 = 1.2,

c2 = 0.5), presented in Figure 5.33. NOx reduction in the non-isothermal condition is

simulated by the model III for the SCR system and using calibrated parameters present-

ed in Chapter 5. The results are illustrated in Figure 5.32 and Figure 5.34 for two differ-

0

100

200

300

400

500

600

0

100

200

300

400

500

600

0 50 100 150 200 250 300 350 400

T (º

C)

NO

& N

H3

(ppm

)

t (s)

NOin

NH3,inNOout

Tout

NH3,out

Tin

Page 112: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 5. Calibration procedure and results 97

ent strategies of ammonia dosages. Ammonia adsorption capacity of the catalyst de-

creases by increasing the temperature, therefore ammonia desorbs from the surface. In

the contrary, rate of NOx reduction increases in high temperature and consequently

more ammonia is consumed in the system. In parallel of sthese two opposite effects,

ammonia dosage increases in the inlet to provide enough ammonia for reduction at high

temperature. While the ammonia feed is increased, amplification of the ammonia and

the two other facts result the ammonia slip rise and after that the outlet ammonia de-

creases. NO conversion is monotonously increases by increasing the temperature. NOx

reduction in the non-isothermal condition is predicted well by numerical simulation, as

shown in Figure 5.32.

Figure 5.33. Calculated heat source at the wall over peripheral area of the cells, for one cell (left

axis) and for entire the monolith (right axis) in two dosages of ammonia α = (0.8, 0.1) respect to the

optimum dosage (αOD for 10 ppm slip).

Figure 5.34 represents the temperature step response of the system for the lower ammo-

nia dosage of 0.8αOD (optimum dosage resulting in 10 ppm slip). Comparing to the

measurements, the increase and decrease in ammonia slip and outlet NO is simulated

well in this condition and the maximum ammonia slip and final conversation of NO is

accurately calculated.

0246810121416

0

20

40

60

80

100

120

140

0 100 200 300 400 500

av (q

wal

l) to

tal [

kW/m

2 ]

av (q

wal

l) pe

r si

ngle

cha

nnel

[W

/m2 ]

t (s)

α = 0.8α = 1.0

Page 113: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

98 Chapter 5. Calibration procedure and results

Figure 5.34. Comparison of predicted and experimental system response to the inlet temperature

step at GHSV = 50,000 h-1, NO/NOx = 100% and ammonia dosage ratios α = 0.8 (with respect to the optimum dosage αOD, resulting in 10 ppm slip). Symbols: experiment, lines: simulation, dashed line:

inlet, red: temperature, blue: NO, green: NH3.

External heat source at the wall is calculated for both cases and as described in eq. (5.2),

the external heat at the wall changes along the channel proportionally to the temperature

difference. Figure 5.33 illustrates the average of the calculated external source over the

wall during the transient operation. The coil heats up the channel in the beginning of the

operation and the external power decreases when the inlet and outlet temperature is

equal and the wall temperature reaches to the maximum value. The calculated heat

sources based on the numerical results for the temperature are almost the same in both

cases. Optimum dosage of ammonia (α=1.0) results efficient NOx reduction and there-

fore the heat released of the exothermic SCR reactions is higher than the lower dosage

of α=0.8. Consequently the required external heat source is lower to warm up the sys-

tem with optimum ammonia dosage. Since the simulation is carried out for a single

channel, the resulted heat is also relative to one channel as presented in the left axis.

Subsequently, the total external heat source for the monolith is extended by considering

the entire cells (9x13) in the sample, as shown in the right axis. Thermal properties of

the monolith are chosen according to the data from the manufacturer such as density of

0

100

200

300

400

500

600

0

100

200

300

400

500

600

0 50 100 150 200 250 300 350 400

T (º

C)

NO

& N

H3

(ppm

)

t (s)

NOin

NH3,inNOout

Tout

NH3,out

Tin

Page 114: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 5. Calibration procedure and results 99

2,500 kg/m3 and thermal conductivity of 1 W/mK. The heat capacity of the monolith is

a function of temperature selected from [78].

The difference in the trend of NOx conversion is illustrated in Figure 5.35 for different

ammonia dosage in the same temperature step. It is seen that the outlet temperature is

not affected by ammonia and the lower dosage of ammonia results smaller ammonia

slip during the observation. NOx reduction at temperature lower than 350 ºC is the same

for both cases; however at high temperature more ammonia results better conversion of

NOx. Ammonia oxidation increases at high temperature therefore there is not enough

ammonia to react with NOx in the case of insufficient ammonia dosage (α = 0.8).

Figure 5.35. Measured response of the system to the inlet temperature step for two different ammo-

nia dosage ratios α = 1.0 and 0.8 (with respect to the optimum dosage αOD, resulting in 10 ppm slip)

at GHSV = 50,000 h-1, NO/NOx = 100% and. Dot-line: outlet in αOD, continues-lines: outlet in

0.8αOD, dashed-line: inlet in αOD, dashed-dot-line: inlet in 0.8αOD, red: temperature, blue: NO,

green: NH3.

0

100

200

300

400

500

600

0

100

200

300

400

500

600

0 100 200 300 400 500 600

T (º

C)

NO

& N

H3

(ppm

)

t (s)

NOin

NH3,out (α=1.0)

NH3,out

(α=0.8)

NOout (α=0.8)

NOout (α=1.0)

Tin

NH3,in (α=0.8)

NH3,in (α=1.0)

Tout

Page 115: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

100 Chapter 5. Calibration procedure and results

5.7 Conclusions

NOx reduction in an NH3-SCR system over a Fe-Zeolite catalyst is simulated by the se-

lected 1D+1D model for the catalytic converter. Three different surface kinetics are

considered in the model and the predicted results are presented in different conditions

and compared by the experimental data. The most recently proposed kinetics consider-

ing dual-site on the catalytic surface with finite rate for the ammonia spill-over reaction

produce the best prediction, for the transient, steady state and ammonia inhibition effect,

and is selected to simulate the system.

An efficient best practice strategy is proposed to parameterize all the parameters needed

for the simulation of the SCR system, section 5.3. The optimization procedure is split

into several steps to reduce the number of parameters to be determined simultaneously.

To efficiently calibrate the model parameters, evolutionary optimization algorithms are

employed. Investigation of the proposed procedure concludes that this calibration meth-

od is ready to employ for any research or commercial Zeolite catalyst used for SCR ap-

plications.

Over the entire temperature range (200 - 450 ºC), for three different NO/NO2 ratios and

varying ammonia dosage at two different GHSVs resulting in a total of 90 operating

conditions, very good agreement was found concerning the transient response of the

system, including effects due to ammonia inhibition, as well as with respect to the cu-

mulative NOx reduction and the ammonia slip when the system reaches steady state,

section 5.4.

Simulation of the dynamic operation of the SCR system, section 5.5, and verification of

the results by experimental data in different conditions (totally 90 cases of space veloci-

ties, temperatures and concentrations) show the strong capability of this model to pre-

dict the performance of the SCR system for highly transient conditions representative of

automotive Diesel engine operation. In conjunction with the successful validation of the

model also for transient system responses reaching steady state in section 5.4 , it can be

Page 116: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 5. Calibration procedure and results 101

concluded, that the model constitutes a robust and comprehensive tool for simulation of

the SCR system for automotive applications.

At non-isotherm conditions, only a preliminary validation could be carried out due to

very limited availability of experimental data. A comparison of the predictions nonethe-

less demonstrates the applicability of the model to predict also responses of the system

under real operating conditions of the Diesel engines with time-varying temperature and

composition of the exhaust gas.

In summary it can be stated that by comparison of the results of the numerical simula-

tion with the experimental data, the calibrated parameters and the simulation method

can be verified. It proves that the model accurately represents the thermo-physical pro-

cess in channels with heterogeneous surface chemistry and demonstrates excellent pre-

dictive capabilities over a broad range of conditions.

Page 117: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 6. Discussion and outlook

6.1 Discussion

Tightening NOx emission standards obligates new reduction methods and the Selective

Catalytic Reduction (SCR) of NOx by ammonia is a promising technique for the abate-

ment of NOx from the exhaust of Diesel engines. Numerical simulation is a powerful

method to speed up the complex design process and development of an SCR system.

Accordingly the 1D+1D model considering the convective mass transfer in the channel

and the intra-porous diffusion within the washcoat layer on the channel walls has been

developed for a single channel of an SCR converter with three different surface chemis-

tries, including a set of the most recent kinetics proposed for Fe-Zeolite catalysts.

The parameters describing the physical and chemical characteristics of the system have

been calibrated based on an extensive experimental dataset of a commercially available

Fe-BEA Zeolite catalyst. The experiments covered a broad range of transient operating

conditions of Diesel vehicles with SCR systems and included also steady state condi-

tions at different space velocities, temperatures and concentrations of NOx and ammo-

nia. The system parameters have been calibrated with the experimental data and the

model has been checked for different temperatures and gas compositions, which are typ-

ical for Diesel vehicles. Different kinetics of the SCR system have been assessed and

the results have been compare in different condition. Consequently the most accurate

and comprehensive kinetics are selected which was recently proposed for Fe-Zeolite

catalysts.

The design of the experimental data allowed the optimization procedure to be split into

five sub-steps as follows, thereby substantially reducing the number of parameters to be

determined simultaneously.

Page 118: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 6. Discussion and outlook 103

• Calibration of adsorption/desorption related parameters in the absence of NOx in

the inlet,

• Calibration of NO/NO2 equilibrium and oxidation parameters in the absence of

NH3 in the inlet,

• Calibration of ammonia spill-over and a sub-set of the SCR parameters at a ref-

erence temperature of 200 ºC,

• Calibration of the remaining SCR parameters, which are active at higher temper-

atures,

• Calibration of a set of reactions completing the SCR process.

At the end of each step, the determined parameters are fixed and remain constant during

the calibration of the remaining parameters in the following steps. No developments

concerning the surface kinetics have been proposed, however the influence of three dif-

ferent rate expressions for the Standard SCR reaction taken from the literature has been

assessed.

The own developed FORTRAN code employed the 1D+1D model for the catalytic con-

verter and the dual-site mechanism is considered for the SCR reactions with finite rate

for the ammonia spill-over reaction. The response of the SCR system is simulated in

totally 180 cases for two different operations: a transient case which reached to steady

state and a very fast (dynamic) operation. Each case was studied at five different tem-

peratures over the range of 200 - 450 ºC for three different NO/NO2 ratios and varying

ammonia dosage at two different GHSVs.

The predicted responses of the system showed an excellent agreement with measured

data in all 180 cases for steady states and during the transitions including effects due to

ammonia inhibition. NOx conversion resulted in steady states and the cumulative NOx

reduction and the ammonia slip during the transition were in a very good agreement

with experiments and simulated accurately by the model.

Page 119: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

104 Chapter 6. Discussion and outlook

In agreement with findings reported in the literature, it was seen that higher concentra-

tions of NO2 enhanced the SCR conversion performance at low temperature due to the

Fast SCR reaction, while addition of NO2 had a negligible impact at high temperature

when the Standard SCR was fast enough. It was further observed that NOx reduction

improved when increasing the temperature, but at the highest temperature a trend rever-

sal occurred. While the residence time is shorter and the ammonia desorption is acceler-

ated at higher temperatures, it was seen that the NOx reduction was mainly controlled by

the ammonia feed at these conditions, for which small differences existed since the ab-

solute values depend on the optimum dosage. It was shown that the optimum dosage of

ammonia (resulted 10 ppm slip) has the best overall performance and more ammonia

dosage did not significantly improve the NOx removal efficiency. The results further

indicated, that a lower space velocity resulted in a better DeNOx performance, which is

in agreement with findings reported in e.g. [54, 79].

Accurate predictions for a broad range of Diesel operating conditions provides strong

evidence of the predictive quality of the developed model and the parameterization of

the chemical kinetics employed. The model can hence be considered a reliable tool for

predicting an SCR system under the highly transient conditions representative of auto-

motive Diesel engine operation.

6.2 Outlook

A real SCR system can be simulated by a multi channel model for the whole converter

which can be coupled by the urea spray simulation in the upstream to model the entire

system. Distribution of ammonia in the entrance of the converter can be calculated by a

spray simulation for aqueous solutions. The model should also consider evaporation,

thermo-hydrolysis and mixing of urea solution. By knowing the ammonia distribution at

the catalyst inlet cross section, a number of representative channels is selected to con-

sider the mixture distribution in the inlet. The presented 1D+1D code can be linked to

the spray simulation software and executed for several channels in parallel. Boundary

Page 120: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Chapter 6. Discussion and outlook 105

conditions at the catalytic wall can be different to consider the interaction between the

channels in the converter.

The simulation of a real system accompanied by experimental data on full scale setup

can be used for non-isotherm transient operation and calibration of the thermal proper-

ties of the systems, since in the laboratory scale the converter is too small and its ther-

mal behavior is negligible comparing the setup and connected facilities. The full scale

setup can be run first artificially by injecting gaseous ammonia to avoid the upstream

simulation. In this case 1D+1D model can be used for multiple channels to find the

temperature distribution.

The complete simulation of an engine test bench can be performed by a multi-channel

non-isothermal model coupled with the upstream urea spray simulation. With such a

simulation setup, predictions of the performance for a full scale SCR system at real op-

erating conditions can be envisioned.

Since the calibration of the system parameters is of industrial interest, a practical and

more automatic procedure can be proposed. An automatic algorithm can be developed

to execute the several proposed steps for calibration of the catalyst parameters, which

will be useful for every research or industrial group who works on similar systems.

Page 121: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären
Page 122: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Nomenclature

a Rate parameter [-]

b Rate parameter [-]

cp,eff Effective heat capacity of wall [J/kgK]

cp,g Heat capacity of gas [J/kgK]

cpsi Cell per square inch

k Rate coefficient [-]

k0 Pre-exponential factor [-]

keff Effective conductivity of wall [W/mK]

kg Gas conductivity [W/mK]

kh Convective heat transfer coefficient [W/m2K]

km,i Mass transfer coefficient [m/s]

nx Number of cells along the channel

ny Number of cells perpendicular to the wall

pO2 Oxygen mole fraction [-]

qH Heat production in the wall [W/m2]

qr Heat of reactions [W/m3]

qwall Heat flux at wall boundary condition [W/m2]

r Formation rate of species [mol/m3.s]

rr Reaction rate [mol/m3.s]

t Time [s]

u Average velocity [m/s]

x Longitudinal direction [m]

xs Non-dimensional axial distance [-]

Page 123: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

108 Nomenclature

sx′ Non dimensional axial distance [-]

y, z, r, n Perpendicular direction [m]

CFL Stability condition

Ci Species concentrations in the channel [mol/m3]

iC

Species concentration inside the pores [mol/m3]

Deff Effective diffusivity [m2/s]

Dg Gas diffusivity [m2/s]

Dh Hydraulic diameter [m]

DPF Diesel Particulate Filter

E Activation energy [J/mol]

EGR Exhaust Gas Recirculation

GHSV Gas hourly space velocity [hr-1]

Hads Heat of adsorption [J/mol]

KLH Equilibrium constant of spill-over [-]

KNH3 Rate constant [-]

KO2 Rate constant [-]

eqK

Semi-equilibrium constant [-]

MR Modified Redox kinetics

NOx Oxides of Nitrogen

NSCR Non-Selective Catalytic Reduction

Pe Peclet number

PM Particulate Matter

R Ideal gas constant [m3.Pa/K.mol]

SCR Selective Catalytic Reduction

T Temperature [K]

Page 124: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Nomenclature 109

T

Wall temperature [K]

Tcoil Coil temperature [K]

Tmax Max temperature [K]

TWC Three Way Catalyst

X Mole fraction [ppm]

* Catalyst acidic site

• Catalyst redox site

Greek letters

Ω Ammonia storage capacity [mol/m3]

Ωc Ammonia storage capacity per mass of catalyst [mol/kg]

α Ammonia dosage ratio, NH3,in/NOx,in [-]

αOD Optimum dosage ratio of NH3,in/NOx,in resulting 10 ppm ammonia slip [-]

β Rate parameter [-]

δ Wall thickness [m]

ε Washcoat porosity [-]

γ Rate parameter [-]

κ Rate parameter [-]

θ Ammonia surface coverage on acidic site [-]

ρeff Effective density of wall [kg/m3]

ρg Gas density [kg/m3]

ω Ammonia surface coverage [-]

Page 125: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären
Page 126: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Referred publications

Three publications have resulted from this study: The models I and II explained in sec-

tions 5.3.3 and 5.4.1 have been published as an ASME paper IMECE2010-40431 [80].

The model III explained in section 5.3.4 has been presented in SAE technical paper

2011-01-2084, JSAE20119195 [81] along with the results for the transient operation

and steady states from section 5.4.2. Model validation for highly transient simulation

using model III presented in section 5.5 has been submitted to the International journal

of engine research [82]. The literature reviews presented in Chapter 1 are based on the

related publication with some minor changes for adaptation to the structure of the PhD

thesis.

Page 127: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären
Page 128: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

References

[1] K. B. Schnelle and C. A. Brown, "NOx Control," in Air Pollution Control Technology Handbook, ed: CRC Press, 2001, pp. 1-15.

[2] A. Raczynski and R. T. Watson, Pollution prevention and abatement handbook 1998: Toward Cleaner Production: The world bank, 1999.

[3] "National primary and secondary ambient air quality standards," Code of Federal Regulations (CFR), vol. 40, 1990.

[4] H. B. Singh, "Reactive nitrogen in the troposphere," Environmental Science & Technology, vol. 21, pp. 320-327, 1987.

[5] "Regulation (EC)," The European parliament and the council, vol. No 595/2009, 2009.

[6] "Clean-Fuel vehicles," Code of Federal Regulations (CFR), vol. 40, 2011.

[7] http://europa.eu/legislation_summaries. (accessed in April 2011). Summaries of EU legislation.

[8] www.epa.gov. (accessed in April 2011). United States Environmental Protection Agency (EPA).

[9] P. Estoueig, "Future Global Heavy Duty Specifications," in 2nd International Conference Lubricants Russia, Moscow, 2006.

[10] G. S. Madia, "Measures to enhance the NOx conversion in urea-SCR systems for automotive applications," Ph.D. thesis, vol. Diss. ETH Zurich No. 14595, 2002.

[11] www.integer-research.com. (accessed in April 2011).

[12] T. V. Johnson, "Diesel Emission Control in Review," SAE Technical Papers, vol. 2008-01-0069, 2008.

[13] T. V. Johnson, "Diesel Emission Control in Review," SAE Technical Papers, vol. 2009-01-0121, 2009.

Page 129: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

114 References

[14] T. V. Johnson, "Review of Diesel Emissions and Control," SAE Technical Papers, vol. 2010-01-0301, 2010.

[15] T. V. Johnson, "Diesel Emissions in Review," SAE Technical Papers, vol. 2011-01-0304, 2011.

[16] L. Landong, et al., "Selective Catalytic Reduction of Nitrogen Oxides from Exhaust of Lean Burn Engine over In-Situ Synthesized Cu−ZSM -5/Cordierite," Environmental Science & Technology, vol. 39, pp. 2841-2847, 2005.

[17] T. V. Johnson, "Diesel Emission Control in Review," SAE Technical paper, vol. 2006-01-0030, 2006.

[18] A. Munnannur and Z. G. Liu, "Development and Validation of a Predictive Model for DEF Injection and Urea Decomposition in Mobile SCR DeNOx Systems," SAE Technical Papers, vol. 2010-01-0889, 2010.

[19] R. Zhan, et al., "Development of a Novel Device to Improve Urea Evaporation, Mixing and Distribution to Enhance SCR Performance," SAE Technical Papers, vol. 2010-01-1185, 2010.

[20] A. R. Ehab, et al., "Modeling of the Injection and Decomposition Processes of Urea-Water-Solution Spray in Automotive SCR Systems," SAE Technical Paper, vol. 2011-01-1317, 2011.

[21] T. L. McKinley and A. G. Alleyne, "A Urea Decomposition Modeling Framework for SCR Systems," SAE Technical Paper, vol. 2009-01-1269, 2009.

[22] B. Scarnegie, et al., "Recent DPF/SCR results targeting US2007 and Euro 4/5 HD emissions," SAE Technical paper, vol. 2003-01-0774, 2003.

[23] W. Müller, et al., "Selective catalytic reduction - Europe's NOx reduction technology," SAE Technical Paper vol. 2003-01-2304, 2003.

[24] D. Chatterjee, et al., "Numerical Simulation of NO/NO2/NH3 Reactions on SCR-Catalytic Converters: Model Development and Applications," SAE Technical paper, vol. 2006-01-0468, 2006.

[25] I. Nova, et al., "NH3-NO/NO2 chemistry over V-based catalysts and its role in the mechanism of the Fast SCR reaction," Catalysis Today, vol. 114, pp. 3-12, 2006.

Page 130: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

References 115

[26] V. I. Pârvulescu, et al., "Catalytic removal of NO," Catalysis Today, vol. 46, pp. 233-316, 1998.

[27] J. L. Williams, "Monolith structures, materials, properties and uses," Catalysis Today, vol. 69, pp. 3-9, 2001.

[28] M. Colombo, et al., "A comparative study of the NH3-SCR reactions over a Cu-zeolite and a Fe-zeolite catalyst," Catalysis Today, vol. 151, pp. 223-230, 2010.

[29] D. Chatterjee, et al., "Numerical Simulation of Zeolite- and V-Based SCR Catalytic Converters," SAE Technical paper, vol. 2007-01-1136, 2007.

[30] A. Grossale, et al., "The chemistry of the NO/NO2-NH3 "fast" SCR reaction over Fe-ZSM5 investigated by transient reaction analysis," Journal of Catalysis, vol. 256, pp. 312-322, 2008.

[31] A. Grossale, et al., "Study of a Fe-zeolite-based system as NH3-SCR catalyst for diesel exhaust aftertreatment," Catalysis Today, vol. 136, pp. 18-27, 2008.

[32] J. C. Wurzenberger and R. Wanker, "Multi-Scale SCR Modelling, 1D Kinetic Analysis and 3D System Simulation," SAE Technical Papers, vol. 2005-01-0948, 2005.

[33] J. Votruba, et al., "Heat and mass transfer in monolithic honeycomb catalysts-I," Chemical Engineering Science, vol. 30, pp. 117-123, 1975.

[34] R. D. Hawthorn, "Afterburner catalysts effects of heat and mass transfer between gas and catalyst surface," A.I.Ch.E. Symposium Series, vol. 70, 1974.

[35] J. Votruba, et al., "Heat and mass transfer in honeycomb catalysts-II," Chemical Engineering Science, vol. 30, pp. 201-206, 1975.

[36] E. Tronconi and P. Forzatti, "Adequacy of lumped parameter models for SCR reactors with monolith structure," AIChE Journal, vol. 38, pp. 201-210, 1992.

[37] E. Tronconi, et al., "Selective catalytic removal of NOx: a mathematical model for design of catalyst and reactor," Chemical Engineering Science, vol. 47, pp. 2401-2406, 1992.

Page 131: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

116 References

[38] D. Chatterjee, et al., "Numerical Simulation of Ammonia SCR-Catalytic Converters: Model Development and Application," SAE Technical Paper, vol. 2005-01-0965, 2005.

[39] E. Tronconi, et al., "A complete model of scr monolith reactors for the analysis of interacting NOx reduction and SO2 oxidation reactions," Chemical Engineering Science, vol. 49, pp. 4277-4287, 1994.

[40] R. Khodayari and C. U. Ingemar Odenbrand, "Selective catalytic reduction of NOx: a mathematical model for poison accumulation and conversion performance," Chemical Engineering Science, vol. 54, pp. 1775-1785, 1999.

[41] G. Groppi, et al., "A comparison of lumped and distributed models of monolith catalytic combustors," Chemical Engineering Science, vol. 50, pp. 2705-2715, 1995.

[42] E. Tronconi, et al., "Experimental and theoretical investigation of the dynamics of the SCR - DeNOx reaction," Chemical Engineering Science, vol. 51, pp. 2965-2970, 1996.

[43] M. Koebel and M. Elsener, "Selective Catalytic Reduction of NO over Commercial DeNOx Catalysts: Comparison of the Measured and Calculated Performance," Industrial & Engineering Chemistry Research, vol. 37, pp. 327-335, 1998.

[44] N. Mladenov, et al., "Modeling of transport and chemistry in channel flows of automotive catalytic converters," Chemical Engineering Science, vol. 65, pp. 812-826, 2010.

[45] G. Groppi and E. Tronconi, "Theoretical analysis of mass and heat transfer in monolith catalysts with triangular channels," Chemical Engineering Science, vol. 52, pp. 3521-3526, 1997.

[46] B. Roduit, et al., "3-D modeling of SCR of NOx by NH3 on vanadia honeycomb catalysts," AIChE Journal, vol. 44, pp. 2731-2744, 1998.

[47] L. L. Raja, et al., "A critical evaluation of Navier-Stokes, boundary-layer, and plug-flow models of the flow and chemistry in a catalytic-combustion monolith," Catalysis Today, vol. 59, pp. 47-60, 2000.

Page 132: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

References 117

[48] H. J. Chae, et al., "Direct Use of Kinetic Parameters for Modeling and Simulation of a Selective Catalytic Reduction Process," Industrial & Engineering Chemistry Research, vol. 39, pp. 1159-1170, 2000.

[49] G. Ramis, et al., "Fourier transform-infrared study of the adsorption and coadsorption of nitric oxide, nitrogen dioxide and ammonia on vanadia-titania and mechanism of selective catalytic reduction," Applied Catalysis, vol. 64, pp. 259-278, 1990.

[50] J. A. Dumesic, et al., "Kinetics of Selective Catalytic Reduction of Nitric Oxide by Ammonia over Vanadia/Titania," Journal of Catalysis, vol. 163, pp. 409-417, 1996.

[51] L. Lietti, et al., "Dynamics of the SCR-DeNOx reaction by the transient-response method," AIChE Journal, vol. 43, pp. 2559-2570, 1997.

[52] L. Lietti, et al., "Transient kinetic study of the SCR-DeNOx reaction," Catalysis Today, vol. 45, pp. 85-92, 1998.

[53] M. Koebel and M. Elsener, "Selective catalytic reduction of NO over commercial DeNOx-catalysts: experimental determination of kinetic and thermodynamic parameters," Chemical Engineering Science, vol. 53, pp. 657-669, 1998.

[54] C. Ciardelli, et al., "SCR-DeNOx for diesel engine exhaust aftertreatment: unsteady-state kinetic study and monolith reactor modelling," Chemical Engineering Science, vol. 59, pp. 5301-5309, 2004.

[55] M. Koebel, et al., "Enhanced Reoxidation of Vanadia by NO2 in the Fast SCR Reaction," Journal of Catalysis, vol. 209, pp. 159-165, 2002.

[56] E. Tronconi, et al., "Redox features in the catalytic mechanism of the "standard" and "fast" NH3-SCR of NOx over a V-based catalyst investigated by dynamic methods," Journal of Catalysis, vol. 245, pp. 1-10, 2007.

[57] C. Ciardelli, et al., "Reactivity of NO/NO2-NH3 SCR system for diesel exhaust aftertreatment: Identification of the reaction network as a function of temperature and NO2 feed content," Applied Catalysis B: Environmental, vol. 70, pp. 80-90, 2007.

[58] I. Nova, et al., "NH3-SCR of NO over a V-based catalyst: Low-T redox kinetics with NH3 inhibition," AIChE Journal, vol. 52, pp. 3222-3233, 2006.

Page 133: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

118 References

[59] E. Tronconi, et al., "Modelling of an SCR catalytic converter for diesel exhaust after treatment: Dynamic effects at low temperature," Catalysis Today, vol. 105, pp. 529-536, 2005.

[60] C. Ciardelli, et al., "NH3 SCR of NOx for diesel exhausts aftertreatment: role of NO2 in catalytic mechanism, unsteady kinetics and monolith converter modelling," Chemical Engineering Science, vol. 62, pp. 5001-5006, 2007.

[61] O. Kröcher, "Chapter 9 Aspects of catalyst development for mobile urea-SCR systems -- From Vanadia-Titania catalysts to metal-exchanged zeolites," in Studies in Surface Science and Catalysis. vol. Volume 171, P. Granger and V. I. Pârvulescu, Eds., ed: Elsevier, 2007, pp. 261-289.

[62] I. Nova, et al., "Unifying redox kinetics for standard and fast NH3-SCR over a V2O5-WO3/TiO2 catalyst," AIChE Journal, vol. 55, pp. 1514-1529, 2009.

[63] D. Chatterjee, et al., "Modelling of a combined NOx storage and NH3-SCR catalytic system for Diesel exhaust gas aftertreatment," Catalysis Today, vol. 151, pp. 395-409, 2010.

[64] I. Nova, et al., "Transient response method applied to the kinetic analysis of the DeNOx-SCR reaction," Chemical Engineering Science, vol. 56, pp. 1229-1237, 2001.

[65] W. N. Gill and M. S. Suwandi, "Some aspects of isothermal laminar flow reactors," AIChE Journal, vol. 9, pp. 273-277, 1963.

[66] A. Güthenke, et al., "Current status of modeling lean exhaust gas aftertreatment catalysts," in Advances in Chemical Engineering. vol. Volume 33, B. M. Guy, Ed., ed: Academic Press, 2007, pp. 103-283.

[67] M. Devadas, et al., "Influence of NO2 on the selective catalytic reduction of NO with ammonia over Fe-ZSM5," Applied Catalysis B: Environmental, vol. 67, pp. 187-196, 2006.

[68] N. Wilken, et al., "Heat of adsorption for NH3, NO2 and NO on Cu-Beta zeolite using microcalorimeter for NH3 SCR applications," Catalysis Today, vol. 151, pp. 237-243, 2010.

[69] L. Olsson, et al., "Detailed kinetic modeling of NOx adsorption and NO oxidation over Cu-ZSM-5," Applied Catalysis B: Environmental, vol. 87, pp. 200-210, 2009.

Page 134: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

References 119

[70] C. Hirsch, Numerical Computation of Internal and External Flows vol. 1: John Wiley & Sons, 1988.

[71] M. Koebel, et al., "Recent Advances in the Development of UREA-SCR for Automotive Applications," SAE Technical Paper, vol. 2001-01-3625, 2001.

[72] M. Elsener and O. Kröcher, "Parametrisierung Fe-Zeolith," Paul Scherrer Institute (PSI), Switzerland, 2009.

[73] S. Y. Joshi, et al., "On the use of internal mass transfer coefficients in modeling of diffusion and reaction in catalytic monoliths," Chemical Engineering Science, vol. 64, pp. 4976-4991, 2009.

[74] MATLAB, Genetic Algorithm and Direct Search ToolboxTM 2, Users' Guide: The MathWorks, 2004.

[75] S. C. Jung and W. S. Yoon, "Modelling and parametric investigation of NOx reduction by oxidation precatalyst-assisted ammonia-selective catalytic reduction," Proceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering, vol. 223, pp. 1193-1206, 2009.

[76] D. Bhatia, et al., "Experimental and kinetic study of NO oxidation on model Pt catalysts," Journal of Catalysis, vol. 266, pp. 106-119, 2009.

[77] I. P. Kandylas and G. C. Koltsakis, "NO2-Assisted Regeneration of Diesel Particulate Filters:  A Modeling Study," Industrial & Engineering Chemistry Research, vol. 41, pp. 2115-2123, 2002.

[78] C. Depcik and D. Assanis, "One-dimensional automotive catalyst modeling," Progress in Energy and Combustion Science, vol. 31, pp. 308-369, 2005.

[79] D. Tsinoglou and G. Koltsakis, "Modelling of the selective catalytic NOx reduction in diesel exhaust including ammonia storage," Proceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering, vol. 221, pp. 117-133, 2007.

[80] L. Sharifian, et al., "Simulation of NOx Reduction in an Ammonia-SCR System with an Fe-Zeolite Catalyst and Calibration of Related Parameters," ASME Conference Publications vol. IMECE2010-40431, 2010.

[81] L. Sharifian, et al., "Transient simulation of NOx reduction over a Fe-Zeolite catalyst in an NH3-SCR system and study of the performance under different

Page 135: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

120 References

operating conditions," SAE Technical Paper, vol. 2011-01-2084, JSAE20119195, 2011.

[82] L. Sharifian, et al., "Calibration of a model for the selective catalytic reduction with ammonia including NO oxidation and simulation of NOx reduction over an Fe-zeolite catalyst under highly transient conditions," International Journal of Engine Research, vol. revised-submitted, 2011.

Page 136: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Appendix

The results of model III for totally 180 different cases are presented in this chapter

where each figure illustrates three different dosages of ammonia, NH3/NOx = (0.8, 1.0,

1.2) times of the optimum dosage (αOD for 10 ppm ammonia slip). Symbols are experi-

mental data; lines show simulation results and dashed lines are inlet species. The red

color presents ppm of NO, blue is NO2 and NH3 is shown by green.

The different cases include two space velocities, five temperatures, three ratios of

NO/NOx and three dosages of ammonia in two types of transient and dynamic opera-

tions as follow:

• “Transient operation-steady state” and “Highly transient performance (Dynamic

operation)”

• GHSV = 30,000 and 50,000

• T = 200, 250, 300, 350 and 450 ºC

• NO/NOx = 100, 75 and 50%

• α = 0.8, 1.0 and 1.2 times αOD

Page 137: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

122 Appendix

Figure A. 1. Transient operation at GHSV = 50000 h-1, 200 °C, 500 ppm NOx, NO/NOx = 100%

Figure A. 2. Transient operation at GHSV = 50000 h-1, 200 °C, 500 ppm NOx, NO/NOx = 75%

0

50

100

150

200

250

300

200

250

300

350

400

450

500

2000 3000 4000 5000 6000 7000 8000

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

0

100

200

300

400

500

0

50

100

150

200

250

300

350

400

2000 2500 3000 3500 4000 4500 5000 5500 6000

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

Page 138: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Appendix 123

Figure A. 3. Transient operation at GHSV = 50000 h-1, 200 °C, 500 ppm NOx, NO/NOx = 50%

Figure A. 4. Transient operation at GHSV = 50000 h-1, 250 °C, 500 ppm NOx, NO/NOx = 100%

0

100

200

300

400

500

600

700

800

0

50

100

150

200

250

300

6200 6700 7200 7700 8200 8700 9200

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

0

50

100

150

200

250

300

350

400

100

150

200

250

300

350

400

450

500

1000 1500 2000 2500 3000 3500 4000 4500

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

Page 139: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

124 Appendix

Figure A. 5. Transient operation at GHSV = 50000 h-1, 250 °C, 500 ppm NOx, NO/NOx = 75%

Figure A. 6. Transient operation at GHSV = 50000 h-1, 250 °C, 500 ppm NOx, NO/NOx = 50%

0

100

200

300

400

500

0

50

100

150

200

250

300

350

400

1800 2300 2800 3300 3800 4300 4800

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

0

100

200

300

400

500

600

0

50

100

150

200

250

5600 6100 6600 7100 7600 8100

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

Page 140: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Appendix 125

Figure A. 7. Transient operation at GHSV = 50000 h-1, 300 °C, 500 ppm NOx, NO/NOx = 100%

Figure A. 8. Transient operation at GHSV = 50000 h-1, 300 °C, 500 ppm NOx, NO/NOx = 75%

0

100

200

300

400

500

600

0

100

200

300

400

500

800 1300 1800 2300 2800

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

0

100

200

300

400

500

600

700

0

50

100

150

200

250

300

350

400

2000 2500 3000 3500 4000 4500

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

Page 141: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

126 Appendix

Figure A. 9. Transient operation at GHSV = 50000 h-1, 300 °C, 500 ppm NOx, NO/NOx = 50%

Figure A. 10. Transient operation at GHSV = 50000 h-1, 350 °C, 500 ppm NOx, NO/NOx = 100%

0

100

200

300

400

500

600

700

0

50

100

150

200

250

300

4700 5200 5700 6200 6700

NH

3 (pp

m)

NO

& N

O2 (

ppm

)

t (s)

0

100

200

300

400

500

600

700

0

100

200

300

400

500

700 900 1100 1300 1500 1700 1900 2100

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

Page 142: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Appendix 127

Figure A. 11. Transient operation at GHSV = 50000 h-1, 350 °C, 500 ppm NOx, NO/NOx = 75%

Figure A. 12. Transient operation at GHSV = 50000 h-1, 350 °C, 500 ppm NOx, NO/NOx = 50%

0

100

200

300

400

500

600

700

0

50

100

150

200

250

300

350

400

700 900 1100 1300 1500 1700 1900 2100

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

0

100

200

300

400

500

600

700

0

50

100

150

200

250

300

2500 3000 3500 4000

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

Page 143: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

128 Appendix

Figure A. 13. Transient operation at GHSV = 50000 h-1, 450 °C, 500 ppm NOx, NO/NOx = 100%

Figure A. 14. Transient operation at GHSV = 50000 h-1, 450 °C, 500 ppm NOx, NO/NOx = 75%

0

100

200

300

400

500

600

0

100

200

300

400

500

350 550 750 950 1150 1350 1550

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

0

100

200

300

400

500

600

700

0

50

100

150

200

250

300

350

400

300 500 700 900 1100 1300

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

Page 144: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Appendix 129

Figure A. 15. Transient operation at GHSV = 50000 h-1, 450 °C, 500 ppm NOx, NO/NOx = 50%

Figure A. 16. Transient operation at GHSV = 30000 h-1, 200 °C, 500 ppm NOx, NO/NOx = 100%

0

100

200

300

400

500

600

700

800

0

50

100

150

200

250

300

1900 2100 2300 2500 2700 2900 3100 3300

NH

3 (pp

m)

NO

& N

O2 (

ppm

)

t (s)

0

50

100

150

200

250

300

200

250

300

350

400

450

500

2800 3800 4800 5800 6800 7800 8800 9800

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

Page 145: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

130 Appendix

Figure A. 17. Transient operation at GHSV = 30000 h-1, 200 °C, 500 ppm NOx, NO/NOx = 75%

Figure A. 18. Transient operation at GHSV = 30000 h-1, 200 °C, 500 ppm NOx, NO/NOx = 50%

0

100

200

300

400

500

0

50

100

150

200

250

300

350

400

2600 3600 4600 5600 6600 7600 8600

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

0

100

200

300

400

500

600

700

800

0

50

100

150

200

250

300

9100 10100 11100 12100 13100 14100

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

Page 146: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Appendix 131

Figure A. 19. Transient operation at GHSV = 30000 h-1, 250 °C, 500 ppm NOx, NO/NOx = 100%

Figure A. 20. Transient operation at GHSV = 30000 h-1, 250 °C, 500 ppm NOx, NO/NOx = 75%

0

100

200

300

400

500

600

0

100

200

300

400

500

1300 2300 3300 4300 5300

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

0

100

200

300

400

500

600

700

0

50

100

150

200

250

300

350

400

2800 3800 4800 5800 6800 7800

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

Page 147: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

132 Appendix

Figure A. 21. Transient operation at GHSV = 30000 h-1, 250 °C, 500 ppm NOx, NO/NOx = 50%

Figure A. 22. Transient operation at GHSV = 30000 h-1, 300 °C, 500 ppm NOx, NO/NOx = 100%

0

100

200

300

400

500

600

700

800

0

50

100

150

200

250

300

7900 8400 8900 9400 9900 10400 10900 11400

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

0

100

200

300

400

500

600

700

0

100

200

300

400

500

1000 1500 2000 2500 3000 3500 4000

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

Page 148: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Appendix 133

Figure A. 23. Transient operation at GHSV = 30000 h-1, 300 °C, 500 ppm NOx, NO/NOx = 75%

Figure A. 24. Transient operation at GHSV = 30000 h-1, 300 °C, 500 ppm NOx, NO/NOx = 50%

0

100

200

300

400

500

600

700

0

50

100

150

200

250

300

350

400

1200 1700 2200 2700 3200 3700

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

0

100

200

300

400

500

600

700

800

0

50

100

150

200

250

300

4300 4800 5300 5800 6300 6800

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

Page 149: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

134 Appendix

Figure A. 25. Transient operation at GHSV = 30000 h-1, 350 °C, 500 ppm NOx, NO/NOx = 100%

Figure A. 26. Transient operation at GHSV = 30000 h-1, 350 °C, 500 ppm NOx, NO/NOx = 75%

0

100

200

300

400

500

600

700

0

100

200

300

400

500

700 1200 1700 2200

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

0

100

200

300

400

500

600

700

0

50

100

150

200

250

300

350

400

500 1000 1500 2000 2500

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

Page 150: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Appendix 135

Figure A. 27. Transient operation at GHSV = 30000 h-1, 350 °C, 500 ppm NOx, NO/NOx = 50%

Figure A. 28. Transient operation at GHSV = 30000 h-1, 450 °C, 500 ppm NOx, NO/NOx = 100%

0

100

200

300

400

500

600

700

800

0

50

100

150

200

250

300

2800 3300 3800 4300 4800

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

0

100

200

300

400

500

600

700

0

100

200

300

400

500

500 700 900 1100 1300 1500

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

Page 151: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

136 Appendix

Figure A. 29. Transient operation at GHSV = 30000 h-1, 450 °C, 500 ppm NOx, NO/NOx = 75%

Figure A. 30. Transient operation at GHSV = 30000 h-1, 450 °C, 500 ppm NOx, NO/NOx = 50%

0

100

200

300

400

500

600

700

0

50

100

150

200

250

300

350

400

1000 1500 2000 2500

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

0

100

200

300

400

500

600

700

800

0

50

100

150

200

250

300

350

3500 4000 4500 5000

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

Page 152: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Appendix 137

Figure A. 31. Dynamic operation at GHSV = 50000 h-1, 200 °C, 500 ppm NOx, NO/NOx = 100%

Figure A. 32. Dynamic operation at GHSV = 50000 h-1, 200 °C, 500 ppm NOx, NO/NOx = 75%

0

50

100

150

200

0

100

200

300

400

500

0 1000 2000 3000 4000 5000 6000

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

0

100

200

300

400

500

600

0

50

100

150

200

250

300

350

400

6900 7900 8900 9900 10900 11900 12900

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

Page 153: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

138 Appendix

Figure A. 33. Dynamic operation at GHSV = 50000 h-1, 200 °C, 500 ppm NOx, NO/NOx = 50%

Figure A. 34. Dynamic operation at GHSV = 50000 h-1, 250 °C, 500 ppm NOx, NO/NOx = 100%

0

100

200

300

400

500

600

700

0

50

100

150

200

250

13800 14800 15800 16800 17800 18800 19800

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

0

50

100

150

200

250

300

350

400

0

100

200

300

400

500

0 1000 2000 3000 4000 5000 6000

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

Page 154: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Appendix 139

Figure A. 35. Dynamic operation at GHSV = 50000 h-1, 250 °C, 500 ppm NOx, NO/NOx = 75%

Figure A. 36. Dynamic operation at GHSV = 50000 h-1, 250 °C, 500 ppm NOx, NO/NOx = 50%

0

100

200

300

400

500

600

0

50

100

150

200

250

300

350

400

6900 7900 8900 9900 10900 11900 12900

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

0

100

200

300

400

500

600

700

0

50

100

150

200

250

13800 14800 15800 16800 17800 18800 19800

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

Page 155: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

140 Appendix

Figure A. 37. Dynamic operation at GHSV = 50000 h-1, 300 °C, 500 ppm NOx, NO/NOx = 100%

Figure A. 38. Dynamic operation at GHSV = 50000 h-1, 300 °C, 500 ppm NOx, NO/NOx = 75%

0

100

200

300

400

500

600

0

100

200

300

400

500

0 1000 2000 3000 4000 5000 6000

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

0

100

200

300

400

500

600

700

0

50

100

150

200

250

300

350

400

6900 7900 8900 9900 10900 11900 12900

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

Page 156: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Appendix 141

Figure A. 39. Dynamic operation at GHSV = 50000 h-1, 300 °C, 500 ppm NOx, NO/NOx = 50%

Figure A. 40. Dynamic operation at GHSV = 50000 h-1, 350 °C, 500 ppm NOx, NO/NOx = 100%

0

100

200

300

400

500

600

700

0

50

100

150

200

250

13800 14800 15800 16800 17800 18800 19800

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

0

100

200

300

400

500

600

700

0

100

200

300

400

500

0 1000 2000 3000 4000 5000 6000

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

Page 157: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

142 Appendix

Figure A. 41. Dynamic operation at GHSV = 50000 h-1, 350 °C, 500 ppm NOx, NO/NOx = 75%

Figure A. 42. Dynamic operation at GHSV = 50000 h-1, 350 °C, 500 ppm NOx, NO/NOx = 50%

0

100

200

300

400

500

600

700

0

50

100

150

200

250

300

350

400

6900 7900 8900 9900 10900 11900 12900

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

0

100

200

300

400

500

600

700

0

50

100

150

200

250

13800 14800 15800 16800 17800 18800 19800

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

Page 158: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Appendix 143

Figure A. 43. Dynamic operation at GHSV = 50000 h-1, 450 °C, 500 ppm NOx, NO/NOx = 100%

Figure A. 44. Dynamic operation at GHSV = 50000 h-1, 450 °C, 500 ppm NOx, NO/NOx = 75%

0

100

200

300

400

500

600

700

0

100

200

300

400

500

0 1000 2000 3000 4000 5000 6000

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

0

100

200

300

400

500

600

700

0

50

100

150

200

250

300

350

400

6900 7900 8900 9900 10900 11900 12900

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

Page 159: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

144 Appendix

Figure A. 45. Dynamic operation at GHSV = 50000 h-1, 450 °C, 500 ppm NOx, NO/NOx = 50%

Figure A. 46. Dynamic operation at GHSV = 30000 h-1, 200 °C, 500 ppm NOx, NO/NOx = 100%

0

100

200

300

400

500

600

700

0

50

100

150

200

250

300

13800 14800 15800 16800 17800 18800 19800

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

0

100

200

300

400

0

100

200

300

400

500

0 1000 2000 3000 4000 5000 6000

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

Page 160: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Appendix 145

Figure A. 47. Dynamic operation at GHSV = 30000 h-1, 200 °C, 500 ppm NOx, NO/NOx = 75%

Figure A. 48. Dynamic operation at GHSV = 30000 h-1, 200 °C, 500 ppm NOx, NO/NOx = 50%

0

100

200

300

400

500

600

700

0

50

100

150

200

250

300

350

400

6900 7900 8900 9900 10900 11900 12900

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

0

100

200

300

400

500

600

700

0

50

100

150

200

250

300

13800 14800 15800 16800 17800 18800 19800

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

Page 161: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

146 Appendix

Figure A. 49. Dynamic operation at GHSV = 30000 h-1, 250 °C, 500 ppm NOx, NO/NOx = 100%

Figure A. 50. Dynamic operation at GHSV = 30000 h-1, 250 °C, 500 ppm NOx, NO/NOx = 75%

0

100

200

300

400

500

600

0

100

200

300

400

500

0 1000 2000 3000 4000 5000 6000

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

0

100

200

300

400

500

600

700

0

50

100

150

200

250

300

350

400

6900 7900 8900 9900 10900 11900 12900

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

Page 162: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Appendix 147

Figure A. 51. Dynamic operation at GHSV = 30000 h-1, 250 °C, 500 ppm NOx, NO/NOx = 50%

Figure A. 52. Dynamic operation at GHSV = 30000 h-1, 300 °C, 500 ppm NOx, NO/NOx = 100%

0

100

200

300

400

500

600

700

0

50

100

150

200

250

13800 14800 15800 16800 17800 18800 19800

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

0

100

200

300

400

500

600

700

0

100

200

300

400

500

0 1000 2000 3000 4000 5000 6000

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

Page 163: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

148 Appendix

Figure A. 53. Dynamic operation at GHSV = 30000 h-1, 300 °C, 500 ppm NOx, NO/NOx = 75%

Figure A. 54. Dynamic operation at GHSV = 30000 h-1, 300 °C, 500 ppm NOx, NO/NOx = 50%

0

100

200

300

400

500

600

700

0

50

100

150

200

250

300

350

400

6900 7900 8900 9900 10900 11900 12900

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

0

100

200

300

400

500

600

700

0

50

100

150

200

250

13800 14800 15800 16800 17800 18800 19800

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

Page 164: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Appendix 149

Figure A. 55. Dynamic operation at GHSV = 30000 h-1, 350 °C, 500 ppm NOx, NO/NOx = 100%

Figure A. 56. Dynamic operation at GHSV = 30000 h-1, 350 °C, 500 ppm NOx, NO/NOx = 75%

0

100

200

300

400

500

600

700

0

100

200

300

400

500

0 1000 2000 3000 4000 5000 6000

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

0

100

200

300

400

500

600

700

0

50

100

150

200

250

300

350

400

450

6900 7900 8900 9900 10900 11900 12900

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

Page 165: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

150 Appendix

Figure A. 57. Dynamic operation at GHSV = 30000 h-1, 350 °C, 500 ppm NOx, NO/NOx = 50%

Figure A. 58. Dynamic operation at GHSV = 30000 h-1, 450 °C, 500 ppm NOx, NO/NOx = 100%

0

100

200

300

400

500

600

700

0

50

100

150

200

250

13800 14800 15800 16800 17800 18800 19800

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

0

100

200

300

400

500

600

700

0

100

200

300

400

500

0 1000 2000 3000 4000 5000 6000

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

Page 166: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Appendix 151

Figure A. 59. Dynamic operation at GHSV = 30000 h-1, 450 °C, 500 ppm NOx, NO/NOx = 75%

Figure A. 60. Dynamic operation at GHSV = 30000 h-1, 450 °C, 500 ppm NOx, NO/NOx = 50%

0

100

200

300

400

500

600

700

0

50

100

150

200

250

300

350

400

6900 7900 8900 9900 10900 11900 12900

NH

3&

NO

2(p

pm)

NO

(ppm

)

t (s)

0

100

200

300

400

500

600

700

0

50

100

150

200

250

300

13800 14800 15800 16800 17800 18800 19800

NH

3(p

pm)

NO

& N

O2

(ppm

)

t (s)

Page 167: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären
Page 168: Rights / License: Research Collection In Copyright - Non ...4636/eth-4636-02.pdfx-Reduktion wird anhand von d Kinetik rei verschiedene Mechanismen modelliert und die Ergebnisse stationären

Curriculum Vitae

Personal data

Name: Leila Sharifian

Date of birth: 16.September.1980 Karaj, Iran

Nationality: Iranian

Education

2007 - present Ph.D., research scientist and teaching assistant at the Aerothermo-chemistry and Combustion Systems Laboratory of Prof. Dr. K. Boulouchos, Department of Mechanical and Process Engineering, Swiss Federal Institute of Technology (ETH) Zurich, Switzerland

2004-2007 Research scientist “CNG direct injection in internal combustion en-gines”, lecturer and teaching assistant in Mechanical Engineering De-partment, Sharif University of technology, Tehran, Iran

2002 - 2004 M.Sc. in Mechanical Engineering “Optimized Cooling System for PEM Fuel Cell Stack, based on Entropy Generation Minimization”, Sharif University of technology, Tehran, Iran

1998 - 2002 B.Sc. in Mechanical Engineering “Heat Recovery Steam Generator (HRSG) Analysis based on ASME Performance Test Code”, Tehran Polytechnic University, Tehran, Iran

1994 - 1998 Under graduate in Mathematics and Physics, NODET High School, Karaj, Iran

Publications resulted from the doctoral studies

[1] L. Sharifian, Y.M. Wright, K. Boulouchos, M. Elsener, O. Kröcher, "Simulation of NOx Reduction in an Ammonia-SCR System with an Fe-Zeolite Catalyst and Cali-bration of Related Parameters," ASME Conference Publications IMECE2010-40431, 2010.

[2] L. Sharifian, Y.M. Wright, K. Boulouchos, M. Elsener, O. Kröcher, "Transient sim-ulation of NOx reduction over a Fe-Zeolite catalyst in an NH3-SCR system and study of the performance under different operating conditions," SAE Technical Pa-per, 2011-01-2084, JSAE20119195, 2011.

[3] L. Sharifian, Y.M. Wright, K. Boulouchos, M. Elsener, O. Kröcher, "Calibration of a model for the selective catalytic reduction with ammonia including NO oxidation and simulation of NOx reduction over an Fe-zeolite catalyst under highly transient conditions," International Journal of Engine Research, revised-submitted, 2011.