Hydrogen-rich materials as auxiliary reducing agents in the blast furnace

(1)

Hydrogen-rich materials as auxiliary reducing agents in the blast furnace

Dimitrios Sideris

Chemical Engineering, master's level (120 credits) 2018

Luleå University of Technology

Department of Civil, Environmental and Natural Resources Engineering

(2)

1

Abstract

The blast furnace is an energy intensive and efficient counter current heat exchange apparatus used in ironmaking. Energy consumption occurs mainly through usage of fossil fuels and an important effect on the environment is the release of pollutants, with carbon dioxide being its largest airborne emission.

Counter measures to reduce resource utilization and environmental impact of the blast furnace are sought through injection of auxiliary reducing agents. These materials are used in combination with the main reducing agents to increase the overall efficiency of the process while decreasing CO2 emissions.

This project intends to evaluate the use of four hydrogen rich materials as auxiliary reducing agents in the blast furnace. The materials tested in this study are carbonaceous materials that have undergone torrefaction or no preprocessing. The hydrogen content of these materials is comparatively high (5-6 wt%) and the expectancy to mitigate the carbon dioxide emissions by substituting part of the pulverized coal (that is the currently used injection material) is reasonable. At the same time the fact that the materials tested are secondary materials originating from the recycling chain reduces the carbon footprint of the overall process.

Kinetic parameters of the materials’ reactions (devolatilization, gasification and combustion) have been determined, along with the materials’ particle size, true density, calorific value and composition. The interaction of the materials’ ashes with coke substrates has also been investigated in order to acquire insight about the effect of the materials’ residue on coke reactivity and consequently its integrity.

Ultimate goal of these investigations is to apply the data and parameters derived to a Computational Fluid Dynamics model and have a credible estimation about the effect of these materials when injected into the blast furnace, avoiding costly pilot scale experiments and industrial trials.

(3)

2

Acknowledgements

I would like to thank my supervisor Hesham Ahmed (LTU) for his help and guidance throughout the project. I am also most grateful to Caisa Samuelsson (LTU) for giving me the chance to conduct a thesis in the Processmetallurgy laboratory and for all her support. Britt- Louise Holmqvist (LTU) contributed the maximum to the realization of my experiments and for this I thank her. Special thanks to Martin Ölund (Swerea MEFOS) for explaining me basic theory and answering to all my questions.

Finally, I would like to express my humble gratitude to everyone in my Department who provided me with means and motivation to complete this task.

(4)

3

List of Figures

Figure 1.1. Outline of the blast furnace mass balance (Geerdes, et al., 2015) ... 7

Figure 1.2. Zones in the blast furnace (Geerdes, et al., 2015) ... 8

(6)

5

Figure 1.3. Auxiliary reducing agents injection (Geerdes, et al., 2015) ... 9

Figure 1.4. Pulverized coal reaction in the raceway (Ishii, 2000) ... 10

Figure 4.1. Material pretreatment process scheme ... 23

Figure 4.2. Temperature profile during devolatilization ... 26

Figure 4.3. Temperature profile during combustion ... 27

Figure 4.4. Temperature profile during gasification ... 29

Figure 5.1. Automated particle size analysis of the 53-106 μm size fraction ... 32

Figure 5.3. Comparison between experimental and theoretical dry mass HHV ... 35

Figure 5.4. TGA graphs during devolatilization... 36

Figure 5.5. TGA graph during combustion ... 39

Figure 5.6. TGA graph during gasification ... 40

Figure 5.7. Mass spectrometry graphs illustrating the samples' devolatilization ... 42

Figure 5.8. Mass spectrometry graphs illustrating the samples' combustion ... 44

Figure 5.9. Mass spectrometry graphs illustrating the samples' gasification ... 46

Figure 5.10. Comparison of ash production using different techniques ... 48

Figure 5.11. Swelling behavior of PC ash ... 49

Figure 5.12. Swelling behavior of PUR ash ... 50

Figure 5.13. Ternary phase diagram of CaO-SiO2-MgO with fixed 10 wt% Al2O3 (Process Metallurgy Course, 2017) ... 51

Figure 5.14. Isothermal section of the CaO-SiO2-Al2O3 phase diagram at 1800K (MTDATA, 2010) ... 52

Figure 5.15. Comparison of SEM photomicrographs of coke before (left) and coke after (right) thermal treatment ... 52

Figure 5.16. Photomicrograph of PC ash on coke using SEM (498x) ... 53

Figure 5.17. Photomicrograph of PUR ash on coke using SEM (85x)... 53

Figure 5.18. Photomicrograph of Carbon PIMIENTO ash on coke using SEM (999x) ... 54

Figure 5.19. Coke reactivity evolution after thermal treatment with the injection materials' ashes ... 55

Figure 5.20. Photomicrograph of Carbon PODA ash on coke using SEM (201x) ... 55

List of Tables

Table 2.1. Particle reactions during auxiliary reducing agents injection ... 15

Table 3.1. Proximate, ultimate and ash analyses ... 21

Table 4.1. Machine parameters ... 23

Table 4.2. Stream definition ... 23

Table 5.1. Helium pycnometry results ... 33

Table 5.2. Bomb calorimetry results ... 33

Table 5.3. Higher heating value for dry materials ... 34

Table 5.4. HHVd calculated using Gaur and Reed formula ... 34

Table 5.5. Results from graphical evaluation of kinetic parameters for devolatilization ... 38

Table 5.6. Results from graphical evaluation of kinetic parameters for combustion ... 39

Table 5.7. Results from graphical evaluation of kinetic parameters for CO2 gasification ... 40

Table 5.8. Ash production by oxidation at 950^oC ... 47

Table 5.9. Summary of heating microscopy results ... 48

Table 5.10. Mass loss during heating microscopy experiments ... 49

Table 5.11. Reduction of ash composition to four basic components ... 50

(7)

6

Abbreviations and Symbols

Abbreviations

CFD Computational fluid dynamics DIA Dynamic image analysis HHV Higher heating value

HV-TSD High volatile torrefied saw dust LTU Luleå Tekniska Universitet MS Mass spectrometry

PC Pulverized coal

PSD Particle size distribution PUR Polyurethane

RAFT Raceway adiabatic flame temperature SEM Scanning electron microscopy

TGA Thermogravimetric analyzer

Symbols

A Pre-exponential kinetic factor Ap Particle surface area

C<S> Char

d50 50% passing size Dp Char particle diameter E Activation energy

Ea Apparent activation energy

fs Mass fraction of reacting solid species in a particle k Kinetic rate constant

kG Granular model kinetic rate constant

kMV Modified volumetric model kinetic rate constant

m Total mass remaining

mc Mass of remaining char mo Initial sample mass 𝑀_𝑂₂ Oxygen molecular weight mRC Mass of unreacted coal

mVM Mass of remaining volatile matter np Number of particles in a sample 𝑃_𝑂₂ Oxygen partial pressure

P80 80% passing size

pg Bulk partial pressure of reacting gas R Universal gas constant

RC Particle surface reaction rate rcom Combustion rate

RD Diffusion rate coefficient RK Reaction rate coefficient RVM Rate of devolatilization

t Time

T Absolute temperature

Vp,o Initial particle volume Vtotal,o Initial sample volume

X Conversion

𝑋_𝑂₂ Oxygen mole fraction in the gas

ρ Density

φ Ratio of reacting surface to external area

(8)

7

1 Introduction

1.1 Background

The blast furnace constitutes the most efficient way of producing pig iron. The basic principles that govern its function originate from the antiquity but it acquired its current form during the last three centuries.

The main operating principle of the blast furnace is the reduction of iron oxides into metallic iron. This is accomplished by the effective contact of the iron minerals with reducing agents, a reaction that is achieved in several ways (regarding the physicochemical state of the reactants) and under various conditions (temperature, pressure) throughout the furnace’s different zones.

An outline of the blast furnace process is illustrated in the following figure.

Figure 1.1. Outline of the blast furnace mass balance (Geerdes, et al., 2015)

The charge materials or stock consists of iron ore, coke and fluxes. These are solid materials which are fed at the furnace’s charging system on top and slowly travel downwards, undergoing several changes during their descent, ending up being collected as hot metal and liquid slag, at the bottom of the furnace through notches and as gases at the gas uptake on top. The main cause of these physicochemical transformations of the burden is the oxygen injected as air hot blast through the tuyeres. The hot blast creates voidage in front of the tuyeres where coke is consumed by oxidation with oxygen producing CO at elevated temperatures. The resulting gas which is a mixture of the reducing CO gas and the inert gaseous components of the air blast ascends through the furnace melting and reducing the burden and ends up at the gas uptake at the top of the furnace.

Concerning the furnace configuration, it is divided into several zones that can be distinguished from each other because of the different physical and chemical status of the materials flowing

(9)

8 through them, their temperature profile and their position. These zones are formed during the blast furnace operation and are namely the following:

Figure 1.2. Zones in the blast furnace (Geerdes, et al., 2015)

• Throat: this is where the solid materials fall after being fed into the furnace through the charging system. Ore and coke are charged in discrete layers and in the throat they form the stockline, where they are first dried by the ascending off gases and get heated to approximately 200^oC.

• Shaft or stack: in this zone the burden is in the solid state but reacts with the ascending gasses that contain CO and H2 and gets reduced from the higher iron oxides (hematite- Fe2O3, and magnetite-Fe3O4) into the lower iron oxides (wustite-FeO, and iron-Fe), while at the same time gets heated to 1100-1200^oC.

• Belly: this region is occupied by alternate layers of permeable solid coke and impervious, semifused mass of iron and primary slag, through which the ascending gases are unable to flow. It is also called the cohesive zone and the gases diffuse in the burden volume through the coke slits and cause further reduction. The gangue in admixture with the flux starts to fuse in this region at temperatures above 1200^oC.

• Bosh: here the reduction is completed and the ores are melted down. The sectional area of the furnace is reduced by about 20-25% in harmony with the resultant decrease in the apparent volume of the charge. It is at the lower part of this zone where the air blast is introduced through tuyeres, creating a raceway in front of each tuyere where combustion of the coke takes place.

• Hearth: The unburnt coke from the tuyere region descends into the hearth, forming the

‘deadman’ coke layer which saturates with carbon the down coming molten metal. The metal and slag stratify into separate layers in the hearth, from where they are tapped periodically (Geerdes, et al., 2015).

(10)

9

1.2 Auxiliary Reducing Agents Injection

Injection of auxiliary reducing agents in the blast furnace has been practiced for some decades, in order to substitute part of the coke in the process. Auxiliary reducing agents are introduced into the blast furnace through injection with the air blast in the tuyeres, while coke is fed along with ore and fluxes through the top charging system at the blast furnace’s stockline.

Auxiliary reducing agents serve two major purposes: the short term temperature control in the furnace and the reduction of the burden material. In the course of their itinerary in the blast furnace, these materials are injected along with hot air in the tuyeres where they first lose their volatile content through the rapid heating they are subjected to, while the released volatiles react with the atmosphere and combust, thus producing the raceway flame. At the same time the remaining charified solid material undergoes combustion with oxygen while further in the process the char particles gasify with the carbon dioxide formed. The final residue of these processes represents the ash content of the original material and reaches the stagnant coke layer or ascends through the blast furnace along with the high flow of gases interacting with the coke either in the deadman zone or the descending which may alter its properties.

Figure 1.3. Auxiliary reducing agents injection (Geerdes, et al., 2015)

Hydrogen rich materials when co-injected along with the blast generate moisture which provokes the water gas shift reaction in the middle zone of the furnace:

𝐶𝑂 + 𝐻2𝑂 → 𝐶𝑂2+ 𝐻2 R. 1

The hydrogen produced by the water gas shift reaction is more reactive than CO and its reaction with the iron oxides in the middle and upper zones of the furnace produces water, which exits the blast furnace, reducing the final CO2 release (Lundgren, 2013).

1.3 State of the art

The basic types of injection materials at the tuyere level are natural gas, oil and pulverized coal.

Injection of auxiliary reducing agents started in the 1960’s with natural gas in Ukraine, but nowadays the use of pulverized coal is more common, mainly due to price and availability, which are to a large extent influenced by regional factors.

(11)

10 In order to inject pulverized coal in the tuyeres, a plant for processing raw coal has to be installed. This installation has to perform the following processes for preparing coal to be mixed with the air blast in the raceway:

• Grinding

• Drying

• Transportation through the pipelines

• Injection through lances in the blast

When pulverized coal is injected via lances into the tuyeres, it immediately undergoes devolatilization caused by the elevated temperature of the air hot blast. The volatiles that are released ignite and combust by the oxygen in the air blast producing CO2 and H2O, while the remaining solid char particles are also ignited and oxidized by the atmosphere containing O2. In the last step, the remaining char particles reform the generated CO2 and H2O into CO and H2

gas by the carbon solution loss reaction. These steps can be illustrated schematically in Figure 1.4:

Figure 1.4. Pulverized coal reaction in the raceway (Ishii, 2000)

The unburnt charified coal that passes through the raceway boundary enters the coke bed and is consumed along with coke fines in high temperature regions by reaction with CO2 in gas and FeO in slag. As char is more reactive than coke, accumulation of coke fines may occur, causing permeability problems, channeling and low gas efficiency. This is the reason why a high char burnout in the raceway is needed (Ölund, et al., 2017).

(12)

11

1.4 Modelling of auxiliary reducing agents injection through the tuyeres

Injection through the tuyeres is a complex phenomenon since it involves reaction kinetics, mass transfer, heat transfer and momentum transfer. The raceway has to come to a steady or quasi steady state in order for the blast furnace to operate continuously. To be able to combine all these physical and chemical processes in a simulation model that can be used to predict the response of the system to input changes, the first step is to divide the individual phenomena.

The components that constitute the total model are:

• Fluid mechanics: turbulence, particle dispersion

• Particle reactions: devolatilization, char reaction

• Gaseous reactions: homogeneous reactions, turbulent combustion

• Heat transfer: convection, radiation, reaction heat

• Others: pollutant formation, particle deformation, fragmentation, etc. (Ishii, 2000).

1.5 Scope of the current work

The purpose of this project is to test several hydrogen rich carbonaceous materials originating from neutral, renewable carbon sources and/or the recycling chain, in order to assess their suitability of being injected as auxiliary reducing agents into the blast furnace. By using hydrogen rich materials in the process the CO2 emissions of the blast furnace may be reduced while at the same time recycled materials will be used in a profitable manner, thus mitigating the overall carbon footprint of the process.

The way to perform this evaluation is by:

• studying their comminution characteristics and particle size distribution

• studying the composition of the materials (proximate and ultimate analyses)

• performing helium pycnometry in order to derive their true density

• determine the calorific value of the materials by bomb calorimetry

• performing thermogravimetric analysis in order to derive their corresponding reactions kinetic constants and consequently predict their behavior in the raceway.

It has to be noted that the kinetic constants derived in this study are apparent kinetic constants and cannot be compared with reference values for pure compounds, but serve well the purpose of characterizing the materials under investigation in terms of their simulated behavior when injected into the raceway, so as to come to a conclusion which one is more appropriate to improve the performance of the blast furnace.

Another major part of this project is the study of the ash content of the materials under investigation and its interaction with coke when the ash reaches the coke layer after char is gasified. In order to achieve the ash evaluation, a number of experiments were performed, namely:

• Analysis of the ash composition of the materials

• Ash production from the materials by burning the materials in a furnace at 950^oC for 3 hours

• Heating microscopy of ash briquettes on coke substrate to monitor the softening temperature, melting temperature, wettability of molten ash on coke

• Separation of the coke substrate and examination with Scanning Electron Microscopy

(13)

12

• Examination of the reactivity of the evolved coke after contact with the ashes by Thermogravimetric Analysis

The data produced as outcome of these investigations will be later used as input to a Computational Fluid Dynamics simulation software in order to model the behavior of the materials in the blast furnace raceway.

(14)

13

2 Literature review

2.1 Factors influencing injection materials’ behavior in the raceway

2.1.1 Particle size

The common practice for injecting materials into the raceway is to grind and pulverize them, with coal being pulverized to P80=75 μm. By reducing the particle size of the materials, a larger specific surface area is achieved, which facilitates and accelerates devolatilization. This way higher amounts of volatile mater are released from the material, a fact that has an effect on the subsequent CO2 gasification which can be negatively affected (reduced char combustibility and furnace permeability) by the presence of remnant volatile matter in the charified material (Carpenter, 2010).

2.1.2 Composition

Composition of the injection materials in the blast furnace is one of the most decisive factors for their suitability as auxiliary reducing agents. Materials containing high amounts of hydrogen generate less heat in the raceway than materials with higher fixed carbon content but have a high replacement ratio since hydrogen is very efficient in the indirect reduction reaction of iron oxides. The moisture content causes a cooling effect in the raceway due to the endothermic solution loss reaction and injection of moisture increases the reductant rate. The oxygen percentage of the injectants is a material characteristic that lowers the heating value of the injectants since oxidation of the carbonaceous materials cannot take place in case carbon- oxygen bonds have already been formed.

Volatile matter content increases gaseous homogeneous combustion in the raceway since blast oxygen primarily reacts with the injected particles volatile content and can then penetrate into the solid particle’s porous structure to oxidize the solid carbon content. Thus, in case high amounts of VM exist they preferentially consume oxygen leaving the remaining amount for char oxidation, while the overall replacement ratio of the material is low. Volatile matter also has an effect on RAFT, since devolatilization is endothermic.

Sulfur and phosphorus are elements that can degrade the hot metal quality while their removal results in additional costs associated with increased slag volume generation and basicity requirements for sulfur removal or hot metal treatment for phosphorus removal. The sulfur content in coal is preferably below 0.8% while that of phosphorus below 0.05%.

Alkalis can contribute to coke degradation and sinter disintegration while they attack the refractory lining. The way for these effects to take place is by catalyzing the coke gasification reaction and decreasing the coke strength in the lower part of the blast furnace. Furthermore alkali condensation on the lining causes the formation of scaffolds which affects the burden descent and reduces lining life (Lundgren, 2013). The combined upper limit for sodium and potassium oxides is usually 0.1% for coal.

Chlorine is another undesirable element in the injectants composition and if present it exits the blast furnace either through the off gas or the slag. Although generation of dioxins in the blast furnace offgas is not detected, chlorine forms hydrochloric acid which corrodes metal components and in particular steel in the blast furnace gas cleaning system. The limit for coal chlorine is typically 0.05% (Carpenter, 2010).

2.1.3 Calorific value

Calorific value is the heat released during complete combustion of the materials. One of the most important properties of the injectants is the amount of heat generated when oxidized by

(15)

14 the air blast immediately after entering the raceway. This heat is used at the lower part of the furnace to heat up and melt the burden material from where it starts softening (about 1100^oC) to casting temperature of 1500^oC (Geerdes, et al., 2015).

The calorific value of the materials determines the amount of heat that can be supplied to the furnace and does not correspond to the actual heat release of the materials in the raceway, since the overall process in the raceway includes gasification of char thus producing CO and H2. The calorific value provides though an indication about the heat potential of each material and its ability to reduce coke consumption. High calorific value injection materials are expected to increase the heat flux in the raceway and consequently the RAFT.

2.1.4 Ash properties

The ash content of the injection materials along with the composition of the ash play a decisive role in the evaluation of the suitability of a material for injection in the blast furnace. A high ash content of the material can cause lance blockage while it consumes energy to remain in the molten phase and increases the slag volume. At the same time it may contribute to blockage of the raceway through the formation of a ‘bird’s nest’, while its deposition on the stagnant coke layer may alter the reactivity of coke and cause permeability problems in the ‘deadman’ zone.

The composition of the ash content is the major factor influencing its fusion characteristics and an excess of acidic (SiO2) or basic (CaO) oxides may give ash deposition problems due to increased deformation and melting temperatures.

The effect of the injectant’s residue when it comes into contact with coke is of outmost importance because it influences coke reactivity. Most coke weakening by the solution loss reaction takes place in the active coke zone and ashes with high alkali, iron oxides, CaO and MgO content can catalyze the endothermic solution loss reaction in case of effective contact with coke (Björkman, 2017).

2.2 Injectants reactions in the raceway

Reactions between the solid particles and the gaseous atmosphere take place as soon as auxiliary reducing agents are introduced into the gaseous stream. Devolatilization is a process which occurs throughout the solid particle’s volume, while combustion and gasification are surface reactions that take place at the boundary between solid and gas.

Heterogeneous reactions of injected particles with the gaseous atmosphere are highly dependent on temperature, which defines whether the rate limiting step of the overall process is chemical reaction or diffusion of the gaseous reactants and products. In the low temperature range chemical reaction is the rate limiting step, in the middle temperature range the rate is controlled by both chemical reaction and diffusion, while at high temperatures diffusion of reactants and products in the boundary layer limits the rate (Ishii, 2000).

Devolatilization, combustion and gasification are heterogeneous reactions intimately connected to the injected material’s behavior in the raceway. The basic formulas that can describe these phenomena are listed in Table 2.1:

(16)

15

Table 2.1. Particle reactions during auxiliary reducing agents injection

Devolatilization

Raw Injection Materials → Volatile Matter (VM)

→ Char (C<S>) + Residue (Ash) Combustion

C<S> + 0.75 O2 → 0.5 CO + 0.5 CO2

Gasification C<S> + CO2 → 2 CO C<S> + H2O → CO + H2

These reactions cannot describe the process by themselves since they constitute only a part of the overall phenomenon. Moreover, they cannot be separated completely since they overlap in the actual process. However, by studying them separately, insight in the process can be obtained and an injection material’s beneficial and detrimental characteristics can be determined.

2.2.1 Devolatilization

When an auxiliary reducing agent is injected in the blast furnace raceway, it is heated up by convection from the hot blast and radiation from the furnace walls, flame and other burning particles. This causes the material to release gaseous and liquid products which create a burning atmosphere around the particles and further provoke the particles’ devolatilization.

In order to simulate the devolatilization process in mathematical terms, several models can be used. The most primitive is the first order reaction model:

(𝑅𝑎𝑤 𝑚𝑎𝑡𝑒𝑟𝑖𝑎𝑙)→ (𝑉𝑜𝑙𝑎𝑡𝑖𝑙𝑒 𝑀𝑎𝑡𝑡𝑒𝑟) + (𝑅𝑒𝑠𝑖𝑑𝑢𝑒) ^𝑘 ^{R. 2}

where k is the kinetic rate constant.

This model postulates that the rate of devolatilization, 𝑅𝑉𝑀, is proportional to the amount of volatile matter remaining, 𝑚𝑉𝑀:

𝑅_𝑉𝑀=𝑑𝑚_𝑉𝑀

𝑑𝑡 = 𝑘 ∙ 𝑚_𝑉𝑀 Eq. 2.1

where 𝑑𝑚_𝑉𝑀 is the mass change of volatile matter, 𝑑𝑡 is the change in time. The kinetic constant k is defined by the Arrhenius law:

𝑘 = 𝐴 ∙ 𝑒⁻^𝑅𝑇^𝐸 ^{Eq. 2.2}

(17)

16 where A is the pre-exponential factor, E is the activation energy, R is the universal gas constant and T is the absolute temperature.

The competing rate model assumes that devolatilization can be described by a pair of competing first order reactions with corresponding kinetic rates k1 and k2, that control the devolatilization rate over different temperature ranges:

(𝑅𝑎𝑤 𝑚𝑎𝑡𝑒𝑟𝑖𝑎𝑙) <

𝑘2→ 𝑎₂(𝑉𝑜𝑙𝑎𝑡𝑖𝑙𝑒)+(1−𝑎₂)(𝑅𝑒𝑠𝑖𝑑𝑢𝑒)

𝑘1→ 𝑎₁(𝑉𝑜𝑙𝑎𝑡𝑖𝑙𝑒)+(1−𝑎1)(𝑅𝑒𝑠𝑖𝑑𝑢𝑒) R. 3

The expression that describes the releasing rate of volatile matter in this case is:

𝑑𝑚_𝑉𝑀

𝑑𝑡 = (𝑎₁𝑘₁+ 𝑎₂𝑘₂)𝑚_𝑅𝐶 ^{Eq. 2.3}

where ai, ki and mRC are the stoichiometric coefficient, reaction rate constant and mass of unreacted coal in a coal particle (in case injection material is pulverized coal) respectively. The rate constants k1 and k2 are given by Arrhenius type equations and k2 contains a higher activation energy (Ishii, 2000).

2.2.2 Combustion

Combustion takes place at the charified materials surface, in combination with the evaporation of the remnant volatile mater and its combustion in the gas phase. Combustion of the solid material takes place at higher temperatures than devolatilization, so combustion is preceded by devolatilization.

Efforts to model the char combustion near or at atmospheric pressure have produced several results which establish the temperature and oxygen concentration dependence of the process considering single step or multi step reactions, with the corresponding number of kinetic constants. All models presented below assume surface reaction, so C stands for active carbon site.

The Global Power- Law Kinetics model considers the following reaction between the active carbon site and oxygen:

C + O2

→ CO/CO𝑘 2 R. 4

with the corresponding rate law given by:

r_com=kP_Oⁿ₂ _{Eq. 2.4}

(18)

17 where rcom is the combustion rate, k is the kinetic constant and 𝑃_𝑂₂is the oxygen partial pressure.

The Langmuir-Hinshelwood-form model considers the intermediate complex C(O) generation between an active site and an absorbed oxygen atom, which influences the overall kinetic rate according to the reaction mechanism:

2C+O₂→ 2C(O) ^k¹ ^{R. 5}

C(O)^k→ CO ² ^{R. 6}

with the corresponding reaction rate:

r_com= k₁k₂P_O₂

k₁P_O₂+k₂ ^{Eq. 2.5}

The Three-Steps Semiglobal Kinetics model includes two intermediate reactions whose rate depends on their corresponding kinetic constants according to the reaction mechanism:

C+O₂^k→ C(O) ¹ ^{R. 7}

C(O) + O₂^k→ CO − CO² ₂+ C(O) ^{R. 8}

C(O)^k→ CO ³ ^{R. 9}

with the corresponding reaction rate law:

r_com=k₁k₂P_O₂+k₁k₃P_O₂ k₁P_O₂+k₃

2

; CO CO₂= k₃

k₂P_O₂ ^{Eq. 2.6}

The Baum and Street model assumes that the char particles are spherical and that the reaction rate is determined by the chemical and/or diffusion kinetics. This model is expressed by the following equation:

𝑑𝑚

𝑑𝑡 = −𝜋𝐷𝑝2𝜌𝑅𝑇 (𝑋_𝑂₂ 𝑀_𝑂₂) (1

𝑅_𝐷+ 1 𝑅_𝐾)

−1

Eq. 2.7

where dm/dt is the rate of char mass loss, Dp is the char particle diameter, ρ is the coal density, XO2 is the oxygen mole fraction, 𝑀_𝑂₂ is the molecular weight of oxygen, while R is the universal gas constant. RD and RK are the diffusion and reaction rate coefficients respectively, with RK

defined as:

(19)

18

𝑅_𝐾= 𝐴𝜑𝑒⁻^𝐸𝑅𝑇 ^𝛼 Eq. 2.8

where A is the pre-exponential factor, φ is the ratio of reacting surface to external (equivalent sphere) area of the particle and Ea is the chemical reaction activation energy (Barranco, et al., 2009).

Similar to the Baum and Street model is the Multiple Surface Reaction model, where the particle surface reaction rate is controlled by the kinetic rate, RK, and the diffusion rate, RD, according to the formula:

𝑅_𝐶 = 𝐴_𝑝𝑓_𝑠𝑝_𝑔 𝑅_𝐾𝑅_𝐷

𝑅_𝐾+ 𝑅_𝐷 ^{Eq. 2.9}

where Ap is the particle surface area, fs is the mass fraction of reacting solid species in a particle and pg is the bulk partial pressure of reacting gas species (Ölund, et al., 2017).

2.2.3 Gasification

Gasification of injection material chars with CO2 starts in the raceway when the CO2 content of the gaseous atmosphere and the prevailing temperature are adequate for the reaction to occur.

The gasification reaction continues to take place outside the raceway boundaries, where unburnt char fines are entrained into the gas flow. In general the reaction of char carbon with CO2 is slower than combustion and this is reflected in the comparison between the combustion and gasification kinetic parameters.

Several models have been proposed to describe the char CO2-gasification, with the most appropriate to fit the TGA data those that consider a single step reaction. This reaction is the solution loss or Boudouard reaction given by the formula:

C+CO₂→ 2CO ^k R. 10

The simplest model is the Volumetric model, which assumes homogeneous reaction of the char by uniform diffusion of the gas in the entire particle volume. This model can be represented by the formula:

dX

dt=k(1-X) ^{Eq. 2.10}

where X is the material conversion, t is the time and k is the kinetic constant. X is given by the formula:

(20)

19 𝑋 =𝑤₀− 𝑤_𝑡

𝑤₀− 𝑤_𝑓 ^{Eq. 2.11}

where w0 is the weight before gasification, wf the weight after gasification and wt the weight at time t. By integrating the formula in Eq. 2.10, the following expression for the conversion degree is derived:

ln (1 − X) = kt ^{Eq. 2.12}

with k following the Arrhenius law.

k=Aexp (- E

RT) ^{Eq. 2.13}

where A is the pre-exponential factor and E is the activation energy.

The Modified Volumetric model is a variation of the Volumetric model, with the addition of the assumption that the kinetic constant (k) is changing with conversion (X) as the reaction proceeds. The reaction rate and the conversion degree correspond to the following equations:

dX

dt=k_MV(X)(1-X) ^{Eq. 2.14}

and after integration:

− ln(1 − X) = at^b ^{Eq. 2.15}

where kMV(X) is the model corresponding kinetic constant and a and b empirical constants. The kinetic constant can be expressed through the following formula:

k_MV(X) = a¹^bb[−ln (1 − X)]^b−1^b ^{Eq. 2.16}

The Granular model assumes that the reaction occurs at the external surface of the spherical particle and as the reaction moves towards smaller particle diameters, only the ash layer remains. This model is given by the formula:

(21)

20 dX

dt=k_G(1-X)²³ ^{Eq. 2.17}

and the integrated form by:

3[1-(1-X)]¹³=k_Gt ^{Eq. 2.18}

where kG is given by the Arrhenius law (Irfan, et al., 2011).

The multiple surface reaction model also applies for gasification, where the particle surface reaction rate is controlled by the kinetic rate, RK, and the diffusion rate, RD, according to the formula:

𝑅𝐶 = 𝐴𝑝𝑓𝑠𝑝𝑔

𝑅_𝐾𝑅_𝐷

𝑅_𝐾+ 𝑅_𝐷 ^{Eq. 2.19}

where Ap is the particle surface area, fs is the mass fraction of reacting solid species in a particle and pg is the bulk partial pressure of reacting gas species (Ölund, et al., 2017).

(22)

21

3 Materials

The materials under investigation in this project were four hydrogen rich carbonaceous materials:

• High volatile torrefied saw dust (HV-TSD)

• Torrefied food residue with code name Carbon PIMIENTO

• Torrefied food residue with code name Carbon PODA

• Recycled foam of Polyurethane (PUR)

The initial shape of the materials was irregular while they contained some coarse particles greater than 1 cm in size, with the exception of PUR that was already fine in size and granular.

The materials were characterized by means of proximate, ultimate and ash analyses. These analyses were conducted in ALS Scandinavia AB laboratories in Luleå, while the reference PC used in this project has been characterized previously by (Ölund, et al., 2017) and the results are given in Table 3.1:

Table 3.1. Proximate, ultimate and ash analyses

HV-TSD Carbon

PIMIENTO

Carbon

PODA PUR PC

Proximate Analysis (wt%)

Moisture 2.5 2.6 1.8 14.3 1.2

Volatile Matter 72.3 62.7 67.5 66.4 18.4

Fixed Carbon 24.7 12.8 14.7 6.5 69.6

Ash 0.5 21.9 16.0 12.9 10.8

Ultimate Analysis (wt% dry basis)

C 55 48.8 49.7 63.2 79.12

H 5.8 4.7 5.4 6 3.93

N <0.10 2.66 1.2 5.81 1.96

O 38.7 20.5 27 9.5 4.02

Cl <0.01 0.32 0.18 0.43 0.00

S <0.012 0.465 0.212 0.051 0.27

Ash 0.5 22.6 16.3 15.0 10.70

Ash Analysis (wt% in total ash content)

Al 0.24 0.64 2.25 2.28 13.08

Ba 0.44 0.02 0.03 0.95 0.00

Ca 22.60 27.52 19.39 5.30 5.14

Cr 0.01 0.00 0.01 0.15 -

Fe 1.14 0.74 1.57 31.53 5.21

K 9.12 2.99 3.23 0.36 1.30

Mg 2.42 2.61 1.52 1.26 1.74

Mn 2.68 0.06 0.04 0.12 -

Na 0.33 0.84 0.62 0.55 0.29

P 0.96 1.43 0.98 0.12 0.34

S 0.00 3.08 1.69 0.00 -

(23)

22

Si 0.86 2.97 10.86 0.00 24.56

Ti 0.02 0.06 0.17 1.26 -

Others 59.18 57.04 57.65 56.12 48.33

All materials contain a high amount of volatile matter that varies between 63 and 72%, except PC. HV-TSD is the material that contains the lowest amount of ash (0.5%) while it contains the greatest percentage of volatile matter (72.3%). Carbon PIMIENTO and Carbon PODA contain large amounts of calcium, magnesium and silicon, but this is expected since they are food residues. Polyurethane foam is the only material that was not subjected to thermal pretreatment and contains the highest percentage of moisture and the lowest percentage of fixed carbon, although its dry basis carbon content is the highest (63.2%). Polyurethane foam also contains the highest amount of chlorine (0.43% in total solids, mainly due to flame retardant additives) which could pose a problem for utilization through combustion (chloride content can cause corrosion of the steel in the blast furnace gas cleaning system) while a large amount of iron is detected in its ash content that is connected to its origin which for the present project remains unknown. The possibility that PUR could act as a credit material for hot metal production is reasonable since a previous experience with injecting in-plant fines has shown that injected iron oxides are quite early reduced to a state between wustite (FeO) and metallic iron (Björkman, 2017), although the quantity of iron contained in PUR can only have a negligible contribution to the total metallic iron production.

(24)

23

4 Methods

4.1 Material Pretreatment

The materials provided had to undergo comminution and stratification, in order to be suitable for use in the subsequent analytical methods. Comminution was performed using a mortar mill (PULVERISETTE 2), while stratification was done by means of a stack consisting of sieves with nominal aperture sizes 106 and 53 μm and a bottom container. A simplified representation of the pretreatment flowsheet using MODSIM software is illustrated in the following figure:

Figure 4.1. Material pretreatment process scheme

The corresponding machine definitions and parameters are listed in Table 4.1:

Table 4.1. Machine parameters

Machine number Definition

1 Mortar mill

2 Screen (nominal aperture: 106 μm)

3 Screen (nominal aperture: 53 μm)

4 Mixer

Mortar mill is represented by machine number 1 due to availability of shapes in MODSIM, while the sieve stack is analyzed in machines 2 and 3. The materials were processed in the mortar mill for 5 min (except from PUR which was already fine in size) and then sieved in a sieve shaker for 5 min using the configuration described in Figure 4.1.

Table 4.2. Stream definition

Stream number Stream definition

1 Circuit feed

2 Mortar mill feed

3 Mortar mill product

4 Screen 2 oversize

5 Screen 2 undersize

6 Screen 3 oversize

7 Screen 3 undersize

(25)

24 The pretreatment process for each material was carried out until more than 5 g in stream number 6 were collected, so as to have sufficient quantity for the subsequent experiments. Stream number 6 (size fraction 53-106 μm) was used for thermogravimetric analysis and particle size analysis, stream number 7 (size fraction <53μm) was used for density determination while the bulk samples were used for proximate and ultimate analysis, calorific value determination and ash production.

4.2 Particle size analysis

The particle size distribution of a material affects its physicochemical properties, such as the flow characteristics, heat transfer and reactivity. In order for the particle size analysis to be reliable and accurate, the sample analyzed has to be representative of the bulk material. Particle size analysis is usually performed by sieving, but automatic analysis devices based on technologies such as high definition image processing are becoming most common.

Dynamic Image Analysis (DIA) is a method used to automatically measure the particle size distribution of a sample. The operating principle of the DIA method is that the particles of the sample under investigation pass in front of two bright, pulsed led light sources, where their shadows are captured with two digital cameras and analyzed to produce their size distribution curves in real time.

A Retsch CAMSIZER X2 was used to automatically analyze the 53-106 μm samples and produce their Particle Size Distribution graphs employing the Dynamic Image Analysis technology. This apparatus is optimized for fine samples analysis (from 0.8 μm to 8 mm) and the particular samples fall into this category and are hence suitable for analysis with the specific equipment (HORIBA, 1996-2018).

4.3 Helium pycnometry

A helium pycnometer calculates the true volume of a solid from the measured drop in pressure when a known amount of gas is allowed to expand into a chamber containing the sample. This volume, combined with the mass of the sample under investigation, gives the true density of the sample.

An AccuPyc II 1340 helium pycnometer was used to measure the true density of the <53μm sieved samples. The true density of the sieved fine fraction is equal to the true density of the other size fractions, since the true density should not be affected by milling or sieving of the material.

4.4 Bomb calorimetry

A bomb calorimeter consists of a steel container (bomb) where a weighted mass of the sample under investigation is loaded and then the whole inner chamber is pressurized with excess pure oxygen at 30 bar. The bomb is submerged under a known volume of water and the weighted reactant is ignited. The energy released by the combustion as heat crosses the stainless steel wall raising the temperature of the surrounding water jacket. The temperature change in the water is then accurately measured and used to calculate the energy given out by the sample burn.

An IKA C200 bomb calorimeter was used to assess the higher heating value of the samples provided. Approximately 0.5 g of each bulk sample was used for each test, while each material was tested three times (with the exception of PUR that was tested twice) in order to get an average value for each material that would be more credible.

(26)

25

4.5 Thermogravimetric analysis

Thermogravimetric analysis (TGA) is a technique in which the mass of a substance is monitored as a function of temperature or time as the sample is subjected to a controlled temperature program in a controlled atmosphere.

During a thermogravimetric analysis, the sample under investigation is put in a crucible which is supported by a precision balance. This crucible resides in a water cooled furnace and is heated or cooled during the experiment. The mass of the sample is monitored throughout the experiment while a purge gas controls the sample’s environment. This gas may be inert or reactive, flows over the sample and exits through an exhaust (PerkinElmer, Inc., 2010).

Themogravimetric analysis of carbon containing substances can indicate mainly four material characteristics:

• Drying: occurs when moisture and other solvents are removed from the material through evaporation at temperatures around the water boiling point.

• Devolatilization: occurs when loosely bonded hydrocarbon compounds are liberated from the material through heating in an inert atmosphere forming charified residue.

• Combustion under oxygen rich atmosphere: occurs when char undergoes oxidation (complete or partial) with oxygen.

• Gasification under CO2 atmosphere: occurs when char reacts with available carbon dioxide according to the Boudouard reaction and forms gaseous products.

The employed device was a Netzsch STA 409 instrument with simultaneous thermogravimetric measurement (TGA) with sensitivity ±1 μg and differential thermal analysis (DTA) coupled with a quadruple mass spectrometer.

4.5.1 Devolatilization

4.5.1.1 Experimental procedure

Devolatilization was carried out twice for each material, once followed by combustion and once followed by gasification. A weighted amount of ~50 mg of the 53-106 μm sieved fraction of material was used in each experiment. The material was placed into an alumina crucible inside the TG chamber and heated under an argon stream of 100 ml/min with a heating rate of 5 K/min from ambient temperature up to 800ôC. Then the sample was cooled with a cooling rate of 20 K/min up to the starting temperature for the subsequent program, which was 100ôC in case combustion followed or 500ôC in case gasification followed.

(27)

26

Figure 4.2. Temperature profile during devolatilization

4.5.1.2 Modeling methodology

In order to extract the kinetic constants for the devolatilization process, a first order reaction model was selected. By combining the rate equation for volatile matter (Eq. 2.1) and the Arrhenius equation for the kinetic constant (Eq. 2.2), the following formula is derived:

𝑑𝑚_𝑉𝑀

𝑑𝑡 = A · exp (− 𝐸

𝑅𝑇) ∙ 𝑚𝑉𝑀 Eq. 4.1

And by transformation:

ln (dm_VM dt

1

m_VM) = ln(𝐴) − E RT

Eq. 4.2

Consequently by using the mass loss data derived from the experiments and by plotting the first part of the equation against 1/T, a straight line is derived whose extrapolation to the y-axis gives the ln(𝐴) value, while its slope equals to -Ea/R. The value for dmVM/dt (the devolatilization rate) is acquired by dividing the mass difference by the time difference between two subsequent data points.

4.5.2 Combustion

Combustion was carried out once for each material, each time preceded by devolatilization, thus it was implemented on char. Combustion was accomplished by injecting a flow of 200 ml/min of synthetic air (20.9 %O2, 79.1 %N2 by vol.) into the TG chamber where the sample lingered after devolatilization. In order for combustion not to start immediately with the injection of synthetic air and to derive a mass loss curve amenable to analysis, the charified sample was first cooled down to 100^oC before air was injected. Then, under the synthetic air

Combustion starting point Gasification starting point

0 100 200 300 400 500 600 700 800 900

0 50 100 150 200 250

Temperature [oC]

Time [min]

Devolatilization temperature profile

(28)

27 flow, the sample was heated to 700^oC with a heating rate of 2 K/min and then cooled to 200^oC with a cooling rate of 20 K/min.

Figure 4.3. Temperature profile during combustion

In order to model combustion a surface reaction model had to be chosen, where the mass loss depends on particle density and diameter. Thus the Baum and Street model was deemed appropriate, a model customized to spherical char particles combustion. In the specific conditions under which the experiments were performed, the diffusion rate coefficient (RD) was assumed to be much higher than the reaction rate coefficient (RK), so the rate equation (after incorporating the reaction rate coefficient formula) reduces to:

dm

dt = −πD_p²ρRT (X_O₂

M_O₂) (𝐴φe⁻^RT^E^a) _{Eq. 4.3}

Assuming homogeneous composition in the particle, the ratio of reacting surface to external surface of the particle (φ) can be approximated by the ratio between the remnant combustible mass of the char (mc) to the remnant mass of the sample (combustible + ash, m). After transformation, the equation becomes:

ln (dm dt

m m_c

−M_O₂

XO₂πDp2ρRT) = ln(𝐴) −E_a RT

Eq. 4.4

Regarding the particles diameter, Dp, spherical particles with initial diameter equal to the d50 of the measured Particle Size Distribution were assumed. Thus the initial volume of each particle (Vp,o) is equal to:

0 100 200 300 400 500 600 700 800

0 50 100 150 200 250 300 350

Temperature [oC]

Combustion time [min]

Combustion temperature profile

(29)

28 𝑉_𝑝,𝑜 =𝜋𝑑₅₀³

6

Eq. 4.5

and the initial total volume of the sample:

𝑉_{𝑡𝑜𝑡𝑎𝑙,𝑜} = 𝑛_𝑝𝑉_𝑝,𝑜 ^{Eq. 4.6}

where np is the total number of particles in the sample. The initial number of particles (np) was calculated by taking into account the initial sample weight (mo) and the true density (ρ) (measured by helium pycnometry), according to the formula:

𝜌 = 𝑚_𝑜

𝑉_{𝑡𝑜𝑡𝑎𝑙,𝑜} ↔ 𝑛_𝑝𝜋𝑑₅₀³ 6 =𝑚_𝑜

𝜌 ↔ 𝑛_𝑝= 6𝑚_𝑜

𝜋𝑑₅₀³ 𝜌 ^{Eq. 4.7}

Thus the number of particles for each sample can be calculated by using the initial sample weight and the d50 derived by DIA. The number of particles is assumed to remain constant throughout the devolatilization and combustion process (no particle fragmentation is supposed to occur) and so does the true density. For this reason, the particle diameter can be calculated at any time using the sample mass (m) according to the formula:

𝐷_𝑝= ( 6𝑚 𝜋𝑛_𝑝𝜌)

1

3 Eq. 4.8

By substituting Eq. 4.11 into Eq. 4.4, the following formula is derived:

𝑙𝑛 (−𝑑𝑚 𝑑𝑡

1 𝑚_𝑐

𝑀_𝑜₂ X_O₂𝑅𝑇(𝑛_𝑝

6)²³(𝑚

𝜋𝜌)¹³) = ln(𝐴) − Ea

RT

Eq. 4.9

Using the mass loss data and plotting the first part of the equation against 1/T, one gets a straight line whose extrapolation to the y-axis gives the ln (𝐴) value, while its slope equals to -Ea/R.

4.5.3 Gasification

Gasification was performed by injecting a flow of 200 ml/min pure CO2 in the TG chamber after devolatilization was completed and the charified samples were cooled down to 500^oC.

Under the CO2 flow, the samples were heated to 1000^oC with a heating rate of 2 K/min and then cooled to 200^oC with a cooling rate of 20 K/min.

(30)

29

Figure 4.4. Temperature profile during gasification

A heterogeneous surface reaction model where the reaction rate depends on the particle surface, the fraction of reacting solid species and the bulk partial pressure of the reacting gas species was chosen to represent the gasification process. Thus, the gasification TG results were analyzed using the surface particle reaction model. In the temperature range the experiments were conducted, diffusion is much quicker than reaction so the diffusion rate coefficient (RD) is much greater than the reaction rate coefficient (RK). Under the aforementioned assumption and after substituting RK with the Arrhenius equation, Eq. 2.19 reduces to:

𝑅_𝐶 = 𝐴_𝑝𝑓_𝑠𝑝_𝑔𝐴𝑒⁻^𝑅𝑇^𝐸^𝑎 ^{Eq. 4.10}

fs can be substituted by mc/m (in this case mc represents the remaining mass available for gasification). Regarding the particles’ surface area (Ap), spherical particles with initial diameter equal to the d50 of the measured Particle Size Distribution was assumed. This way an initial number of particles (np) was calculated by taking into account the initial sample weight (mo) and the true density (ρ) (measured by helium pycnometry), according to the formula:

ρ= mo

V_total ↔ n_pπd₅₀³ 6 =mo

ρ ↔ n_p= 6mo

πd₅₀³ ρ ^{Eq. 4.11}

The number of particles is assumed to remain constant throughout the devolatilization and gasification process (no particle fragmentation is supposed to occur) and so does the true density. For this reason, the particle diameter can be calculated at any time using the sample mass (m) according to the formula:

0 200 400 600 800 1000 1200

0 50 100 150 200 250 300 350

Temperature [oC]

Gasification time [min]

Gasification temperature profile

(31)

30 𝐷_𝑝= ( 6𝑚

𝜋𝑛_𝑝𝜌)

1

3 Eq. 4.12

while the total surface area of the particles becomes:

A_p=n_pπd²=(n_pπ)¹³(6m

ρ )²³ ^{Eq. 4.13}

By substituting fs and Ap, Eq. 4.10 becomes:

−𝑑𝑚

𝑑𝑡 = (n_pπ)¹³(6m ρ )²³𝑚_𝑐

𝑚𝑝_𝑔𝐴𝑒⁻^𝑅𝑇^𝐸^𝑎 ↔

ln (−𝑑𝑚

𝑑𝑡 (𝑛_𝑝𝜋)¹³(6 𝜌)

−2 3 𝑚¹³

𝑚_𝑐𝑝_𝑔) = 𝑙𝑛(𝐴) −𝐸_𝑎

𝑅𝑇 Eq. 4.14

By plotting the left side of the equation against 1/T one gets a straight line whose extrapolation to the y-axis gives the ln (𝐴) value, while its slope equals to -Ea/R.

4.6 Mass Spectrometry

Mass spectrometry is an analytical technique that detects the substances that compose a sample by separating them according to their molecular mass. The way to achieve this separation is by ionizing a small amount of the material and uniformly accelerating them through a pair of oppositely charged plates. Then a vertical magnetic field deflects the accelerated ions and causes them to follow different trajectories depending on the inertia of each ion which is proportional to its mass-to-charge ratio (Reusch, 2013).

A Quadruple Mass Spectrometer was integrated in the Netzsch STA 409 off gas port in order to monitor the composition of the evolved gases in the Thermogravimetric chamber during the mass loss cycles of the materials under investigation. This way additional information about the evolved gas composition under thermal treatment of the sample would be provided.

4.7 Ash production

Ash was prepared by heating the samples for 3 hours at 950^oC in a muffle furnace where air was allowed to circulate, i.e. the atmosphere was ambient. Approximately 10 g of each of the 4 samples and one pulverized coal reference sample were put in separate crucibles and heated in a muffle furnace under air in order for the volatile and carbon content to oxidize and evaporate. The products of this process (the solid residues) represented the ash content of each material and after cooling were collected in separate bottles.

4.8 Heating microscopy

When a solid material is heated under inert atmosphere, it undergoes phase transitions such as melting, where the ordering forces in the solid lattice disappear and the molecules start to move freely. Transition from the solid to the liquid phase can be observed through a change in external area and form when a test object of the material under investigation is subjected to an appropriate temperature program.

Hydrogen-rich materials as auxiliary reducing agents in the blast furnace