• No results found

Thermodynamic Study of Trace Elements in the Blast Furnace and Basic Oxygen Furnace

N/A
N/A
Protected

Academic year: 2021

Share "Thermodynamic Study of Trace Elements in the Blast Furnace and Basic Oxygen Furnace"

Copied!
65
0
0

Loading.... (view fulltext now)

Full text

(1)

MASTER'S THESIS

Thermodynamic Study of Trace Elements

in the Blast Furnace and Basic Oxygen

Furnace

Anton Andersson

2014

Master of Science in Engineering Technology Sustainable Process Engineering

Luleå University of Technology

(2)

Acknowledgement

This thesis was carried out at Swerea MEFOS and it was the final task before achieving the Master of Science degree in Sustainable Process Engineering. Moreover it serves as a door opener towards future challenges.

I would like to thank my supervisor Mats Brämming at Swerea MEFOS for his input and assistance throughout the work. Also, my supervisor Associate Professor Caisa Samuelsson and examiner Professor Bo Björkman at Luleå University of Technology for their valuable input. Furthermore, a thank you to the engineers at SSAB Luleå for providing process data and other information, Linda Bergman, Magnus Heintz and Anita Wedholm.

Luleå, August 2014 Anton Andersson

(3)
(4)

Abstract

The iron ore based steel producers in Sweden and Finland operates mainly on pellets produced by LKAB. The introduction of new mining sites is believed to influence the future pellet chemistry. Furthermore, environmental and economic factors act as a driving force towards increasing the material- and energy efficiency by increasing the recirculation of in-plant by-products. All together this amounts to a realized change in feed chemistry. Therefore, it is of interest to study how trace elements behave in the process to be able to follow the material flow of trace elements within the integrated steel plant.

In this thesis it was attempted to describe the distribution of trace elements between metal, slag and gas phase in the blast furnace and basic oxygen furnace (BOF) process using thermodynamic equilibrium calculations. The work was focused on developing an approach to calculate the distribution of zinc in the blast furnace and chromium in the BOF. The same approach was then utilized for lead in the blast furnace and cobalt in the BOF to determine if it was applicable on other trace elements as well.

The blast furnace calculations were divided into three different scenarios representing different parts of the furnace; namely, the hearth, thermal reserve zone and the section above the thermal reserve zone. The results showed that for elements having a cyclical behavior in the furnace, such as zinc and lead, the assumed recirculation rate is directly decisive for the calculated output through the tap hole. And, that more data is needed to confidently estimate a probable recirculation rate that fits the calculations. Furthermore, it was shown that, from a thermodynamic standpoint, no lead or zinc leaves the blast furnace through the top. To describe the output of these elements through the off gas it was argued that a thorough study of the connection between furnace operating parameters and the dust and sludge amount and zinc and lead contents of the dust and sludge is required.

The BOF calculations were executed by adding the oxygen in increments to an open system, allowing the gas to leave between each calculation step. The calculations were carried out for and compared to results of a pilot plant scale converter and an industrial scale converter. From the results it was concluded that the distribution of chromium could be described for the pilot plant scale converter although the comparison of the overall composition of the slag and crude steel was not satisfactory. Furthermore, the distribution of chromium for the industrial scale converter could not be described using the method at hand. It was argued that the failure to describe the outcome resided in the fact that thermodynamic calculations were employed on a process where kinetics is known to play an important part. Cobalt could be described using the method. However, a simple mass balance with the assumption that essentially all cobalt reports to the crude steel phase would give the same results.

(5)

Symbols and Nomenclature

The symbols used in this report are explained for below.

𝑘𝑔 𝑡𝐻𝑀⁄ = kg of an element per ton of hot metal, used to describe the load in the blast furnace.

𝑡𝐶𝑆 = ton crude steel

(%𝑀𝑒) = Denotes the weight percentage of component Me in the slag phase.

[%𝑀𝑒] = Denotes the weight percentage of component Me in the molten iron or steel. 𝐵𝑎𝑠𝑖𝑐𝑖𝑡𝑦 (𝐵2) = Ratio between mass percentage of CaO and SiO2 in the slag phase.

𝛾𝑥𝑜 = The Henrian activity coefficient of component x.

𝑋𝑦 = The mol fraction of component y.

𝑇 = The temperature given in Kelvin

(6)

Table of Contents

1 Introduction ... 1

1.1 Background ... 1

1.2 Purpose, Aim and Scope ... 2

2 Literature Survey ... 3

2.1 Blast Furnace Ironmaking ... 3

2.2 Basic Oxygen Furnace Process ... 7

2.3 Modeling of Metallurgical Processes ... 10

3 Methods and Datasets ... 14

3.1 Thermodynamic Data ... 14

3.2 Blast Furnace Calculations ... 15

3.3 Basic Oxygen Furnace Calculations ... 19

4 Results and Discussion ... 23

4.1 Results of the Blast Furnace Calculations ... 23

4.2 Basic Oxygen Furnace... 33

5 Discussion on the Thermodynamic Data used ... 40

6 Concluding Discussion ... 41

6.1 Blast Furnace ... 41

6.2 Basic Oxygen Furnace... 42

7 Conclusions ... 42

8 Further Studies ... 43

9 References ... 44

10 Appendix 1 – Ingoing Values to the Blast Furnace Calculations ... 48

10.1 Hearth Equilibrium Calculations ... 48

10.2 Thermal Reserve Zone Calculations ... 50

10.3 Above the Thermal Reserve Zone and Below the Throat ... 50

11 Appendix 2 – Phase Diagrams ... 52

12 Appendix 3 – Results from Calculations on the 2006 Dataset ... 54

13 Appendix 4 – Results from Basic Oxygen Furnace Calculations ... 55

13.1 Imphos Dataset ... 55

(7)
(8)

1

1 Introduction

The background, purpose and aim of this Master of Science thesis are described below.

1.1 Background

The reason why this thesis is of interest for the industry is briefly described below.

In ore based steelmaking, the blast furnace is the dominating process equipment used for reducing the iron ore to iron. The process operates on coke and iron ore agglomerated as either pellet or sinter. In Sweden and Finland, the iron ore agglomerates used are mainly the LKAB pellets. There are plans for introducing new mining sites which are expected to affect the future chemical composition of the pellet produced.

Furthermore, high primary raw material prices and environmental legislation drive the integrated steel plant to operate at higher material- and energy efficiency. The recirculation of in-plant by-products is a possibility when striving for higher material efficiency and thereby lowering the need of virgin material. The recirculation also reduces the amount of material being landfilled. The by-products subjected for recirculation are internal metal scrap, slags, dusts and sludge from different parts of the process. The slags are utilized to a high extent both as construction material [1] and as slag formers in the process [2]. Most of the iron rich dust and sludge is generally recycled to the sinter in the sinter making process [3]. When operating pellets, some sludge fractions and dusts can be recirculated to the blast furnace through cold bonded agglomeration of briquettes [2].

The dust and sludge from the blast furnace contains high amounts of iron and carbon [3]. The sludge from the basic oxygen furnace (BOF) process contains high amounts of iron [3]. In addition to this, trace elements (with respect to iron and steel) such as zinc, alkalis, lead, antimony, tin and cobalt are present in smaller amounts. Some of these elements such as zinc and alkalis are known to be harmful for the blast furnace process. The knowledge regarding the behavior of the other elements is limited and it is of great interest to investigate how these behave in the process.

At Process Integration at Swerea MEFOS mathematical based models of material flow within the integrated steel plant have been developed. As the demand for an increased recirculation of by-products and a possible change in pellet chemistry may change the chemistry of the feed there is a need of studying the effects of this on the material flow. By thermodynamically studying how trace elements are distributed between the metal, slag and gas phase some insight can be provided to this matter.

(9)

2

1.2 Purpose, Aim and Scope

The purpose and aim of the project is described below. Also, the scope is presented.

The purpose of this project is to study if calculations using thermodynamic equilibrium can be applied to describe the distribution of trace elements between metal, slag and gas phase in the blast furnace and BOF.

The aim is to calculate how zinc is distributed between the hot metal, slag and gas phase in the blast furnace and how chromium is distributed between the crude steel, slag and gas phase in the BOF process. These elements were chosen as there are available thermodynamic data describing the elements and data to use as comparison with the calculations. Furthermore, the project also aims towards using the same way of calculating to investigate if the approach can be applied for the distribution of lead in the blast furnace and cobalt in the BOF.

The thesis is limited to describe zinc in the blast furnace and chromium in the basic oxygen furnace as well as applying the developed calculation approaches to lead in the blast furnace and cobalt in the BOF.

(10)

3

2 Literature Survey

A brief theory regarding iron ore based steelmaking mainly focusing on the blast furnace and basic oxygen furnace process is described below. Also, published research of models describing the blast furnace and basic oxygen furnace are presented.

2.1 Blast Furnace Ironmaking

The production of pig iron from iron ore is described briefly below.

An illustration of the iron blast furnace is presented in Figure 1 below. The raw material used in the process is coke, iron ore agglomerates of pellets or sinter and slag formers. [4] Other materials that can be used are cold bonded briquettes of residues, basic oxygen furnace slag [2] and scrap. Alternating layers of coke, iron burden and slag formers are charged at the top. The material descends due to gravity when the coke is continuously burned at the tuyere level and the slag and pig iron is tapped from the hearth.

In the tuyeres hot blast, i.e. preheated air, is introduced together with pulverized coal. The blast adds to the thermal balance of the blast furnace. Also, the reaction between the blast and the pulverized coal and the coke generates carbon monoxide. As the carbon monoxide ascends throughout the furnace the iron ore consisting mainly of hematite, Fe2O3, is reduced. [4]

When the iron oxide is reduced to metallic iron throughout the descent in the furnace the gangue material in the iron ore is reporting to the slag phase together with ash from the combustion of coke and pulverized coal. The iron is softened and melted in the cohesive zone and trickles together with the slag through the coke layers down to the hearth. Due to the density difference between the slag phase and hot metal (HM) the slag floats on the HM. The two phases can thus be separated when tapping the furnace. [4]

(11)

4

Figure 1: Cross section of the blast furnace. Reactions occurring at different heights are presented. [5]

2.1.1 Reactions and Temperature Profile

The ideal temperature profile and reduction scheme in the blast furnace is described in this section.

The reduction of hematite to metallic iron occurs in different stages with different requirements on the reducing gas. To describe this, the carbon monoxide (CO) utilization factor, %ηCO, is introduced as for Equation 1 below [4]:

%𝜂𝐶𝑂 = 100% ∙% 𝐶𝑂+% 𝐶𝑂% 𝐶𝑂2 2 (1)

Table 1 below presents the different utilization factors and CO/CO2 ratios at

equilibrium for the reduction of the iron oxides at 900 °C. The reduction reactions given in the table are called the indirect reduction of the iron oxides.

Table 1: CO-utilization factors and CO/CO2 ratios at equilibrium for

the reduction of iron oxides at 900 °C. [4]

Reaction CO/CO2 CO

3Fe2O3 + CO = 2Fe3O4 + CO2 0 Ca. 100 Fe3O4 + CO = 3FeO + CO2 0.25 80

(12)

5

From the table it is seen that at 900 °C the required ratio to successfully reduce wüstite, FeO, to metallic iron is 2.3 and the equilibrium is as shown in Reaction 1 below:

𝐹𝑒𝑂 + 3.3 𝐶𝑂 = 𝐹𝑒 + 𝐶𝑂2+ 2.3 𝐶𝑂 (1)

The counter-current outline of the process enables a gas rich in CO to come in contact with the wüstite. The gas is then successively depleted in CO while ascending in the furnace reducing the iron oxides.

An ideal temperature profile is depicted in Figure 2 below. The reactions occurring are also given.

Figure 2: Ideal temperature profile for the blast furnace represented as height above tuyere level. [4]

From the above it is clear that the conditions in the blast furnace with respect to temperature and chemical environment are changing with respect to the position in the vertical direction of the furnace. This is of great importance to consider when estimating the distribution of elements between metal, slag and gas by equilibrium calculations.

(13)

6

2.1.2 Behavior of Zinc in the Blast Furnace

The cyclical behavior of zinc in the blast furnace is described below. Also, thermodynamic studies of zinc in systems of iron, carbon and zinc are accounted for.

2.1.2.1 Zinc Input, Output and Cyclical Behavior

Zinc enters the blast furnace in small quantities through the iron ore as oxides, ferrite, silicates or sulfide [6]. It may also enter the process through the coke, recirculated BOF slag [7] and cold bonded briquettes of by-products [8]. The zinc inputs vary between 0.05 and 2.5 kg/tHM [9] noting that some European producers report inputs below 0.16 kg/tHM [10] [11] [12] [13]. The zinc compounds are reduced to metallic zinc vapor by the CO rich gas in the lower regions of the blast furnace where the temperature exceeds 1000 °C. The zinc vapor follows the ascending gas and is reoxidized to zinc oxide in the cooler parts of the furnace. The condensation occurs at temperatures below 520-580 °C [4] and the fine particles are either deposited on the lining and the burden material or exit the blast furnace through the off gas [6]. The deposition of zinc on the lining may disrupt the bricks due to the volume expansion of the transformation from gaseous to solid state. Other problems are scaffold formation which may disturb the burden descent [4]. The zinc deposited on the burden travels down to the lower region where it is reduced and volatilized again, forming a cyclical behavior [6]. The circulation of zinc within the blast furnace is negative with respect to the consumption of reducing agents [7]. The circulating load of zinc is the highest in the temperature region of 800-1200 °C. Samples have shown that zinc concentrations are ten times higher in the shaft than in the charged burden [9].

The zinc output through the off gas is increased by a large difference in the temperature of the top gas and burden material; i.e. a high top gas temperature and a low burden material temperature [9]. Also, a strong central gas flow is favorable for zinc removal at the top. The zinc content in the off gas was studied with laser induced breakdown spectroscopy in [9] showing a periodical change in zinc output related to the charging mechanism of alternating layers of coke and ferrous burden. It was also shown that a decrease in blast rate was correlated to decrease in zinc output through the off gas.

In addition to the off gas, the zinc may leave the blast furnace through the hot metal and slag phase. The zinc removed by tapping is increased with decreasing flame temperature as well as decreasing silicon and manganese content of the hot metal. However, most of the zinc is assumed to evaporate during the tapping, reporting to the cast house dust [9].

Material balance over the blast furnace has shown that 15-27 % of the zinc exits through the flue dust, 45-70 % through the sludge obtained when wet cleaning the off

(14)

7

gas, 5-10 % through the hot metal, 5 % through the slag. In addition to this 5-10 % is deposited on the refractories inside the furnace [6].

2.1.2.2 Thermodynamic Studies of Zinc

The research available on thermodynamic interpretation of the iron-carbon-zinc system is focused mainly on the steelmaking processes due to the use of secondary raw materials such as galvanized steel scrap. However, the carbon contents studied are close to the levels seen in the blast furnace where after the results are applicable in this context as well.

The activity coefficient of zinc has been studied with two different methods; namely thermochemical equilibration of zinc in an iron-carbon alloy achieved by metal-metal equilibrium [14] or gas-metal equilibrium [15] [16]. All studies present a strong positive deviation from the Raoultian behavior. However, the results contradict each other when considering the behavior with changing temperature. The results presented by [14] suggest an increase in zinc activity coefficient with increasing temperature while the results of [15] suggest a decrease in the zinc activity coefficient. Also, the activity coefficient achieved in [14] is considerably larger than that of [15]. A possible explanation suggested by the authors of [14] was the difference in range of XZn in the studies.

2.1.3 Behavior of Lead in the Blast Furnace

The available information regarding the behavior of lead in the blast furnace is not as thorough as that of zinc. A short description is presented below.

Lead is brought into the process through the ore [4] or limestone in varying amounts between 10 and 50 g/tHM [9]. The lead compounds are completely reduced in the upper part of the furnace [4]. Since lead is insoluble in iron and has higher density than both the hot metal and slag it flows down and accumulates in the hearth. Although metallic lead has a low vapor pressure, some of it may vaporize and condense in the upper part of the furnace on the burden or lining material [4] [9]. It may also leave with the top gas reporting to the dust or sludge fraction [9]. Thus, lead may show a cyclical behavior as that of zinc.

2.2 Basic Oxygen Furnace Process

The production of crude steel in the BOF is described below.

The hot metal from the blast furnace may be pre-treated before entering the basic oxygen furnace to allow for optimal operation when producing high quality steel. The pre-treatment can be designed to remove one or more of the elements silicon, phosphorous and sulfur. [17]

(15)

8

The main purpose of the BOF process is to oxidize the carbon in the steel. The hot metal produced in the blast furnace has carbon content in the range of 4-5 % [17] [18] which is reduced to the desired level, usually below 0.1 %. Other important aspects are control of sulfur and phosphorous as well as reaching the desired temperature of the melt. [17]

2.2.1 Basic Oxygen Furnace Process Outline

Below follows a brief description of the outline of the operation.

Figure 3 depicts the general operational steps of the top-blown BOF process. The distribution between the steel scrap and hot metal is in the range of 75-95 % hot metal and remainder scrap depending on local operational conditions [18]. Additional coolants may be added to utilize the heat evolved during the process. Examples are iron ore, pre-reduced pellets [17] or possibly cold bonded pellets of in-plant by-products [19]. The latter would favor the material efficiency of the integrated steel plant. The composition of the scrap is hard to measure and influence the presence of e.g. elements such as chromium.

Figure 3: General outline of BOF process. [17]

Oxygen is blown through a water cooled lance at supersonic speed into the metal bath under strict control of lance height above the bath level. The oxygen oxidizes iron, silicon, carbon, manganese and phosphorous [17]. Lime and dolomitic lime are added as fluxes. The lance height and flux additions are controlled to reach desired slag formation procedure [18]. The oxidized carbon leaves with the gaseous phase. Oxides of iron, manganese, phosphorous are transferred to the slag phase together with calcium sulfide, CaS [17]. Thus, undesired elements together with iron oxides are

(16)

9

transferred to the slag phase during the decarburization to reach a steel of certain composition.

The steel is sampled to ensure that the desired composition and temperature of the steel have been reached. The slag and metal are tapped separately. [17] The kinetic energy of the tapping is utilized for mixing of alloys.

2.2.2 Change in Metal- and Slag Composition

This section briefly describes the change in metal- and slag composition as well as temperature in the BOF process.

The change in chemical composition during the blowing period is illustrated in Figure 4 below. As the oxidation continues the temperature rises from about 1340 °C to the desired temperature in the range of 1650 °C [18].

Figure 4: Change in chemical composition of the metal phase during the BOF process. [18]

It is clear from Figure 4 that the prerequisites for thermodynamic calculations of trace elements change during the process. There is a difference in both temperature and oxygen activity of the melt during the blow. Also, presence of other elements may alter the activity of the trace elements and thereby the distribution. It can also be noted the difference as compared to the blast furnace which is carried out at reducing conditions and at lower temperatures.

Figure 5 below illustrates the change in slag composition during course of the process. During the oxygen blowing, the slag varies in composition. Some of these effects cannot be accounted for in a thermodynamic calculation. This is true for e.g. the CaO dissolution behavior and FeO formation as these phenomena are not equilibrium controlled.

(17)

10

Figure 5: Change in slag composition during the oxygen blowing. [18]

2.2.3 The Oxidation of Chromium in Steelmaking

This section presents a brief description of the behavior of chromium in steelmaking.

Chromium is dissolved as divalent and trivalent chromium in the slag phase. Increasing temperature, decreasing oxygen potential and decreasing slag basicity results in an increase of the ratio between Cr2+ and Cr3+. Since the BOF process is operated under high slag basicity and high oxygen potential Cr3+ predominates in the slag phase [18].

The chromium distribution ratio, (%Cr)/[%Cr], has been shown to be proportionally increasing with the FeO content of the slag. This is based on data from measurements of slag and metal phases from tapping of the electric arc furnace, laboratory experiments and the outdated open hearth furnace [18].

2.3 Modeling of Metallurgical Processes

This section covers research published in the field of modeling of metallurgical processes such as the blast furnace and the basic oxygen furnace as well as examples from other metallurgical industries.

Pyrometallurgical processes are operated at high temperatures with, in general, high reaction rates. On this basis, the assumption of chemical equilibrium can be utilized to simulate the outcome of the operation. The equilibrium calculations can be performed by utilizing software which calculates the quantities of all species in the different phases by a Gibbs energy minimization routine [20] [21].

(18)

11

To be able to describe processes where the outcome is deviating from equilibrium, mass and heat transfer can be introduced to the model [22] [23] [24] [25]. In these non-equilibrium models the assumption of high reaction rates are assumed to be valid locally. The reactor is therefore divided into several sections. In each section, chemical equilibrium is assumed to be reached and the distribution of the species in the different phases is calculated through a Gibbs energy minimization routine. The material and heat transfer between the sections are defined in accordance with the process conditions.

2.3.1 Modeling of the Blast Furnace

The complexity of the blast furnace process is difficult to model even with all the knowledge and experience that have been acquired throughout the years. Some different approaches and simulations are presented below.

As summarized by [24] there are multiple ways to approach the problem of modeling the blast furnace. Mathematical models, models based on data mining, heat and mass balance models and thermodynamic models are some categories that can be distinguished. Each approach has its pros and cons. In this section, focus will be paid towards models with underlying thermodynamic calculations.

A thermodynamic two-step model of the blast furnace was developed [25] to quantitatively describe the behavior of alkalis in the process. The blast furnace was divided into two sections; namely, a hearth reactor and a gas condenser representing the shaft. The mass and heat flow of the model is illustrated in Figure 6 below. Equilibrium was assumed for both reactors and the calculations were proceeded until the alkalis satisfied a defined mass balance criteria between the input and output. The calculations were made for different scenarios of alkali load, basicity and hearth temperature. The results were consistent with plant observations, however, noting that absolute values provided by the model should not be taken as exact. A suggested improvement for the model was to divide the gas condenser into several steps to more accurately describe the alkali behavior.

Figure 6: Illustration of the mass and heat flow of the two-stage model. [25]

(19)

12

The division of the blast furnace shaft into more sections was realized in the five-unit model presented by [24]. The blast furnace was divided into three equilibrium reactors and two heat exchangers as depicted in Figure 7. The equilibrium reactor, R2, represents the thermal reserve zone of the blast furnace. An interesting assumption made was complete reduction to metallic iron before entering the equilibrium reactor, R1, representing the hearth. In the hearth reactor the final slag and hot metal composition was calculated. The formation of a slag phase was exclusively considered in this reactor, i.e. not in R2. The last reactor, R3, represents the fast heat exchange, in the upper part of the blast furnace, between the burden material and gas phase as seen in Figure 2. The purpose of this reactor was to calculate the final equilibrium composition of the off gas.

The comparison to industrial data was limited by the analysis of the iron ore in the dataset. However, accuracy of the predictions was argued to be good despite slight overestimation of the carbon and silicon content of the hot metal.

Figure 7: The schematics of the five-unit model. R1, R2 and R3 are equilibrium reactors and C1 and H1 are heat exchangers. Explanations of the different streams are found elsewhere. [24]

2.3.2 Modeling of the Basic Oxygen Furnace Process

Research published in the field of simulation of the BOF process is generally focused on the decarburization and dephosphorization. Some models are explained for below.

(20)

13

In BOF steelmaking the carbon and oxygen content of the melt differs from equilibrium [26]. From this standpoint, most of the published research is incorporating measures to account for deviations from equilibrium.

A top-blown basic oxygen furnace was modeled [22] using the non-equilibrium approach. The converter was divided into four steady-state reactors. In each reactor, chemical equilibrium was assumed and calculated by minimizing the Gibbs energy. The reactors represented distinct parts of the converter with correlation to the outline of the process; a hot-spot-, metal-slag-, metal bath reactor and a slag mixer. The model was designed to simulate the outcome for stepwise addition of oxygen to the hot-spot reactor. The model showed good correspondence to published data with regards to the decarburization reaction and silicon content of the liquid metal.

The decarburization and dephosphorization in the BOF process was simulated with a kinetic approach by [27]. The model developed was based on a previous model for hot metal dephosphorization. The reaction kinetics was described by the double-layer theory, explained for in [28], allowing for mass and heat transfer in the model. The model outline is out of the scope of this thesis. However, it is noted that the model accurately simulated the slag FeO content related to the metal carbon content. Also, the simulation results using this model for phosphorous, silicon and manganese were similar to the compared dataset from the industry.

A model describing the BOF composed of a reaction model and a model for material melting and dissolution was developed by [29]. The reaction model was designed to consider both thermodynamics and kinetics. It was based on the assumption that only iron is oxidized by the top blown oxygen. The other elements were then subjected to a coupled oxidation-reduction reaction with the iron oxide. The melting of added scrap was described by two different mechanisms; namely, diffusive scrap melting and forced melting. The latter, where the temperature is above the melting temperature of the material, was also used for the other additions such as ore and FeSi. A validation of the model was presented in [30] comparing simulation results with the outcome from both a 170-ton and 330-ton converter. The calculated metal and slag composition agreed well with the actual. Also, the behavior of the slag and metal during the blow was similar to that presented in section 2.2.2 above.

To model the process phenomena in a top-blown BOF converter, computational fluid dynamics software was coupled with thermodynamic databases [31]. To construct a dynamic model using this approach is out of the scope of this work and such a model is found elsewhere [31].

(21)

14

2.3.3 Modeling of Other Pyrometallurgical Processes

Models including thermodynamic calculations have been proven to be successful in predicting the outcome of other metallurgical processes as well. Some examples are given below.

The process of argon oxygen decarburization (AOD) in stainless steelmaking was simulated by [20] using equilibrium based calculations. The different oxidation stages with altering argon oxygen quotients were simulated individually using the outcome of the previous stage as input in the next stage. For each stage the oxygen and argon gas mix was provided stepwise, allowing equilibrium to be reached after each addition. The model was found to predict the trends of the process well. The accuracy was argued to be lowered due to lack of thermodynamic data, the assumption of constant temperature and from not considering known kinetic factors of the AOD process. An equilibrium based model of a flash converter was developed and evaluated in [21]. The process was simulated by introducing an initial amount of oxygen followed by stepwise addition of oxygen to calculate the distribution of the simulated elements between the phases as the process proceeded. The model predicted the outcome well, suggesting that the process mainly depends on thermodynamic properties.

Modeling of a Pierce Smith converter was performed in [23]. A dynamic, non-equilibrium, model was developed. The predicted values were compared to plant data and equilibrium calculations. The non-equilibrium approach showed better agreement with plant data than only equilibrium calculations. The model was developed by dividing the converter into horizontally aligned sections connected by heat and mass flows. In each section chemical equilibrium was assumed to be reached.

3 Methods and Datasets

The method employed for the calculations are presented below together with the datasets used as comparison and the thermodynamic data utilized.

The calculations were performed with the thermochemical software FactSage 6.1. To calculate the equilibrium composition the module Equilib was used. This function utilizes the Gibbs minimization algorithm and thermochemical functions of ChemSage [32].

3.1 Thermodynamic Data

The thermodynamic data used in the calculations are presented below.

The thermodynamic data used in the calculations were taken from the FactSage databases. For solid oxides, solid oxide solutions and slag phases the FToxide database was used. The solution Aslag-liq was used to describe the slag phase in both the BOF

(22)

15

and blast furnace. For the liquid iron- and steel phase, Fe-LQ from FTmisc was used. Gaseous species were taken from the Fact53 database.

The Henrian activity coefficient for zinc in a liquid iron-carbon-zinc system is given in Equation 2 below. It was taken from the function derived from the experiments conducted in [15].

ln 𝛾𝑍𝑛𝑜 =3490𝑇 − 0.142 + �2136𝑇 + 2.86� 𝑋𝐶 − �3915𝑇 + 1.83� 𝑋𝐶2(2)

The activity coefficient was recalculated for the carbon content reported for hot metal in the blast furnace operation to represent the form shown in Equation 3 below.

log 𝛾 = 𝐴 +𝐵𝑇 (3)

Where A and B are constants. The activity coefficient function was merged with the Fe-LQ solution.

3.2 Blast Furnace Calculations

The approach used for the blast furnace calculations is presented below together with a short description of the datasets used to compare the calculations with.

In this thesis, no software enabling mass and heat transfer between different reactors was used. From section 2.1 it is clear that the blast furnace cannot be covered in one calculation step where thermodynamic equilibrium is assumed to be valid. To account for the difference in chemical environment and temperature the calculations were performed for a different set of scenarios representing the various parts of the furnace; namely, the hearth, the thermal reserve zone and the part above the thermal reserve zone. The methodologies for these are accounted for in the subheadings below.

The datasets used for comparison were two different zinc balances for blast furnace no. 3 at SSAB Luleå. The first balance was over a long term period provided for the entire operation of 2012; this dataset is referred to as the 2012 dataset. The second was a period over three days during which a full-scale trial with lowered top pressure was performed [33]; this dataset is referred to as the 2006 dataset. Operational data from SSAB Luleå was used as the foundation in assumptions regarding temperatures, pressures and %ηCO.

3.2.1 Hearth Equilibrium Calculations

The calculations performed for the lower part of the blast furnace were designed to treat the hearth as an equilibrium reactor. The assumptions regarding the calculations are explained for in detail below.

The hearth equilibrium calculations were performed for both the 2006 and 2012 dataset. The elements considered in the calculations were chosen to represent the

(23)

16

major constituents of the gas, slag and hot metal phase. These elements are Fe, C, Si, Mn, Ca, Mg, Al, O and N together with the trace element Zn. Elements important in other aspects were left out, e.g. hydrogen, sulfur and phosphorous.

It was assumed that all iron ore is completely reduced to iron before entering the hearth. The material inputs, temperature and pressure used in the calculations are accounted for in appendix 1. The inputs were calculated from the material balances for 2012 and 2006 provided by SSAB. In the data from SSAB, for each year the average composition for the different charged raw materials are given. Also, the input amount is given as an average for each month and for the entire year. For the dataset of 2012 the average composition and input for the entire year was used. For 2006 the average composition for the entire year was used together with the average input for July, i.e. the month specific for the full scale trials. For convenience, all calculations were performed on a per ton hot metal basis.

No solid oxide phases were allowed to exist. This was based on that solid particles in the slag phase are unwanted and avoided in real operation due to the increase in slag viscosity. Furthermore, the slag was not allowed to form before the hearth, i.e. the oxides were added as oxides to the calculations and not as a single liquid slag phase. The limestone charged was assumed to be completely burned during its descent down the furnace.

Solid carbon in the form of graphite was allowed to exist to represent the dead man in the hearth. Graphite was chosen instead of coke since there were no data describing the solubility of coke in liquid iron in the databases used. No distinction between coke carbon and carbon from the pulverized coal injection was made. The refractory was neglected. The hearth temperature was assumed to be the same as the tapping temperature. The pressure was assumed to be the same as the blast pressure and all the blast was allowed to be in equilibrium with the hot metal and slag.

Calculations in [25] showed that the circulating load of alkali increases linearly one to one with increasing alkali load. The same was assumed for zinc in these calculations although the behavior regarding zinc output is different. The circulating load of zinc was assumed to be ten times the input based on the information provided by [9].

3.2.1.1 Illustrating the Effect of Assumptions

The effect of assumptions related to blast volume and behavior of zinc was illustrated with calculations based on the 2012 dataset. The reactants, temperature and pressure used in the calculations are presented in appendix 1. In one scenario the zinc was assumed to enter the hearth as condensed zinc oxide. The entire circulating load of zinc was therefore used as input in the calculations regardless of the amount of blast. In the other scenario zinc was assumed to be gaseous. Therefore, a decrease in blast amount resulted in decreased ingoing zinc in the calculations.

(24)

17

To enable the use of different blast volumes for a fixed amount of ingoing solids the reaction between carbon and oxygen was calculated as a stoichiometric combustion to carbon monoxide. The gas phase used as input was therefore composed of CO(g) and N2(g). The carbon left was also introduced in the calculations to allow dissolution into the hot metal.

3.2.1.2 Trends in Zinc Output

A useful application area of thermodynamic calculations is the information that can be received from changed conditions. Therefore, a different set of changes to the 2012 dataset was made to illustrate trends in the behavior of zinc. In [9] there is a summary from Russian blast furnaces which reports circulating loads up to and above 20 kg/tHM. Therefore, the effect of these high loads on the zinc output from the hearth was investigated. The articles referred to in [9] presenting the circulating loads for the Russian blast furnaces could not be accessed to validate how these loads were estimated. However, [33] refers to studies on frozen blast furnaces that have shown circulating loads up to 7.2 kg/tHM. Therefore, it was assumed that the high levels reported by the Russian producers are justified to use in the calculations. Furthermore, scenarios of changes in hearth temperature and pressure were calculated. The input, temperatures and pressures used in the calculations are accounted for in detail in appendix 1.

3.2.2 Thermal Reserve Zone Calculations

The method used to calculate how zinc behaves in the thermal reserve zone is presented below.

The blast furnace at SSAB Luleå is capable of operating with a top pressure up to 1.5 atm over pressure. The span of pressures that the outgoing gas may be subjected to is thus between 1 and 2.5 atm. Normal operating conditions show blast pressures of about 3.5 atm and top pressures of approximately 2 atm. Since the furnace is designed to operate with increased top pressure it was assumed, in the calculations, that the pressure in the shaft may vary between 1.5 and 3.5 atm depending on the vertical position in the furnace and the top pressure.

The input for the calculations is presented in appendix 1; it is based on the 2012 dataset. The hematite charged to the blast furnace is assumed to be reduced to wüstite before entering the thermal reserve zone. The carbon left in the hearth equilibrium calculations is assumed to be available for direct reduction of the wüstite. Cementite, Fe3C, was assumed not to form and carbon formation was not allowed. As argued by

[24], the slag formation in the blast furnace is dependent on iron ore chemical composition and local equilibrium conditions. To avoid this complexity a liquid slag phase was not allowed forming in this part of the furnace.

(25)

18

The gas phase was imported from the hearth equilibrium calculations to represent the ascending gas in the process. Slag formers were added as oxides with the exception of calcium. Part of the calcium input was set as limestone while the calcium in the charged BOF slag and briquette was set as lime.

The retention time of the thermal reserve zone was assumed not to favor the kinetics of metal oxide solid solutions or spinel phases with zinc. Thus, the only forms of zinc allowed was Zn(g), Zn(s), Zn(l) and ZnO(s).

The calculations were performed for two different conditions with respect to temperature and pressure. The first condition was designed to represent operation with high top pressure and a probable temperature. The pressure was assumed to be 3 atm, allowing an assumed pressure drop of about 0.5 atm over the cohesive zone. The temperature was set as 900 °C which is within the temperature interval reported for the thermal reserve zone in the literature. The second calculation was designed to find out at what temperature the zinc gas is oxidized and condensed as zinc oxide. A pressure of 2.75 atm was chosen to represent a mean value over the thermal reserve zone.

3.2.3 Calculations for Above the Thermal Reserve Zone and below the Throat The method for the calculations representing the zone above the thermal reserve zone is described below.

Considering the approach presented above in 3.2.2 it is clear that a major limitation is the inability to allow changes in pressure and gas composition in the gas while ascending through the thermal reserve zone. This was enabled by not introducing the solids in these calculations. The calculations were built up as a gas cooling study with changing conditions in both pressure and %ηCO.

The temperature of the gas leaving the thermal reserve zone was set as 900 °C and was cooled down to 500 °C in the calculations. At gas temperatures below this, the interaction with the burden was assumed to be only heat transfer. Therefore, the %ηCO reached its final value of 54.3 % in the calculations, which is the actual value provided by SSAB. The %ηCO of the gas leaving the thermal reserve zone was calculated using

the results from the hearth equilibrium calculations as base. The carbon left in that calculation was assumed to directly reduce wüstite to iron forming carbon monoxide. The carbon monoxide of the gas leaving the hearth was then assumed to react with the remainder of the wüstite to form iron and carbon dioxide. The resulting gas composition that was used as input in the calculations is presented in appendix 1. Since no solids are considered, the %ηCO value was changed manually. It was assumed

to increase linearly to the its final value during the gas cooling, starting at the value calculated as described above.

(26)

19

As previously mentioned, the blast pressure is about 3.5 atm in operation and a pressure drop of 0.5 atm was assumed over the cohesive zone. From the ideal temperature profile depicted in Figure 2 it can be seen that the thermal reserve zone is roughly estimated to 11 m high and the section above about 4.5 m high. A linear pressure drop was assumed above the cohesive zone which means that the pressure drops from 3 atm to the top pressure of 2 atm in 15.5 m. From this it was calculated to use a pressure decreasing from 2.3 to 2 atm in these calculations.

The calculations were also conducted for top pressures of 1.5 and 1 atm. The same assumption regarding linear pressure drop was utilized here. The blast pressure was assumed to decrease as much as the top pressure. The calculation scheme for all three cases is presented in appendix 1.

3.2.4 Lead Calculations

The approach used for zinc was tried for lead as well. Additional assumptions and alternations to fit the expected behavior of lead are described below.

Due to the behavior of lead as described in section 2.1.3 it was assumed that the load of lead in the hearth is higher than the charged. Comparing the description of zinc and lead in the literature it was assumed that the load was 20 times the input of lead. The calculations for the hearth equilibrium were performed in the same way as for zinc. The input of lead was determined from analyses of trace elements in the pellets, limestone and BOF slag. The lead content of the charged briquette was assumed to be 100 ppm. Analyses of lead in hot metal were provided for different tappings. All analyses showed the same value for lead despite a difference in hot metal composition. Therefore, it was assumed that the lead output from the bottom rather depends on input and load in the hearth instead of hot metal composition. It was therefore decided to perform the lead calculations on the 2012 dataset.

A gas cooling study was performed for the gas phase output from the hearth equilibrium calculations. It was conducted in the same way as for zinc in section 3.2.3 above.

3.3 Basic Oxygen Furnace Calculations

The assumptions and procedure employed when calculating the outcome of heats of the BOF is presented below.

3.3.1 Datasets for Comparison

The datasets used for comparison are accounted for below.

Two datasets were used to compare with the BOF calculations. The first dataset was the Imphos trials [34], a project founded by the European Research fund for Coal and Steel (RFCS). The second was composed of nine heats from the BOF at SSAB Luleå.

(27)

20

The Imphos trials were carried out in a six ton converter with metal and slag sampling before, during and after the blow. Chromium content in both hot metal, crude steel and final slag was provided. Out of the 23 available heats eleven were chosen to represent a wide span in slag FeO content. The exclusion of heats was made based upon poor chromium balances, comments in trial data and uncertainties in reported analyses. The dataset from SSAB Luleå represents the full scale operation as SSAB process over 20 times more crude steel per heat as compared to the Imphos trials. The chromium content of hot metal and crude steel was provided in the dataset. Normally, the slag is not sampled which means that no chromium content of the slag phase was provided. In fact, the slag analyses provided for the heats were all calculated by the software used by SSAB. The heats were chosen to represent a span in calculated slag FeO content.

3.3.2 Calculation Basis

The best results were obtained using two different approaches for the datasets where after these are described in separate headlines. Some general information true for both datasets is presented below.

The elements considered in the calculations were Fe, C, Si, Mn, Cr, Ca, Mg and O. The slag analyses found in [34] shows that vanadium reported as V2O5 may constitute up to six weight percent of the slag weight, i.e. a considerable amount. Vanadium was neglected in the calculations due to lack of thermodynamic data and possible effects of this oxide on the slag properties were lost.

In the BOF process, slag formers are added at carefully chosen times into the oxygen blowing to favor the slag formation. In the calculations all material input was introduced before the oxygen blowing started. Adding the slag formers at a certain time into the blowing period was considered to be unnecessary since the complex slag forming process is not thermodynamically controlled.

There are at least two different approaches available when considering the gas phase in the calculations. One is to add all the oxygen at once in one equilibrium calculation. Another is to add the oxygen stepwise and allow the gas phase to leave after each equilibrium calculation, i.e. an open system. As argued by [20], calculating in an open system will lower the partial pressure of carbon monoxide. This will not have any significant effect at higher carbon contents. At lower carbon contents a higher partial pressure of carbon monoxide will result in higher crude steel carbon content. Since the Imphos dataset consists of charges with carbon content down to 0.02 % it was preferred to use the open system approach. Also, from a process point of view it is more realistic to allow the gas phase to leave. Therefore, the open system approach was used.

(28)

21

During the blow, the temperature steadily increases in the process due to the oxidation of the different elements. Using the open system approach, the software did not allow calculations of a final temperature adiabatically or under heat losses. Therefore, a fix temperature was chosen to apply during the entire blow. The pressure was assumed to be one atmosphere.

The material input for the calculations were calculated from the reported charged weights of hot metal, scrap, lime, dolomite, raw dolomite and iron ore. The composition of the scrap used in the Imphos trials was estimated as a general analysis for scrap of higher quality used at SSAB. Estimations of the composition of different scrap qualities charged to the SSAB process were provided to calculate the amount of ingoing material. These analyses are left out of the report.

Figure 19 and Figure 20 in appendix 2 illustrates that the slag phases may be saturated in e.g. lime and periclase. From the figures it was decided to allow solid CaO, MgO and Ca2SiO4 to exist. It was also assumed that a metal oxide solid solution could exist,

based on solid solutions rich in MgO.

Chromium was presented as Cr2O3 in given slag analyses. In the calculation both CrO and Cr2O3 was allowed to form in the slag phase. However, after the thermodynamic calculation, the CrO was recalculated to Cr2O3 to ease the comparison with the actual values.

3.3.2.1 Assumptions and Method Specific for the Imphos Dataset

For the most part, no MgO was charged in the Imphos trials. Instead, the input of MgO was calculated from dissolved refractory reporting to the outgoing slag.

The ingoing material together with the slag analysis was used to calculate a slag weight. The slag weight was taken as the average for two calculations; one based on silicon and one on calcium. In the dataset no crude steel weight was provided. This was estimated using the calculated slag weight and total charged iron together with the slag and crude steel analysis in a material balance. This means that no losses of iron to the gas phase was assumed. The values from the slag weight and crude steel weight calculations were then incorporated in a chromium balance calculation to estimate the difference between ingoing and outgoing amount of chromium. If the difference was too high the heat was not considered.

The oxygen input to the calculations was calculated based on the stoichiometric oxygen demand for carbon oxidation and the oxidation of elements to the slag. Comparing this to the actual injected amount of oxygen, the remaining oxygen was assumed to go to post combustion of carbon monoxide. This assumption means that carbon dioxide was not allowed to form in the equilibrium calculations. The oxygen was added in 100 steps in an open system as described earlier.

(29)

22

The fix temperature used in the calculations was chosen as the end of blow temperature. For the heats with end temperatures well over 1700 °C the average of the two last temperature measurements of the blow were used instead of the end of blow temperature.

3.3.2.2 Assumptions and Method for the SSAB Dataset

The approach used for the SSAB dataset differs from the Imphos in the way oxygen is added. The oxygen was added in increments of 48.541 kg O2 until the desired carbon content of the crude steel was reached. The amount of each addition was calculated from the assumption that 340 Nm3/min is added to the process. This was recalculated to kg of O2 per six seconds which results in 48.541 kg O2. Six seconds was chosen to

limit the number of calculation steps while keeping the possibility to stop at the desired carbon content.

To avoid problems of no slag formation only CaO and MgO was allowed to be present as solids in the slag phase. The fixed temperature was chosen as the end of blow temperature. Carbon dioxide was not allowed to exist in the gas phase although the oxygen addition in this calculation method differed. This was based on the assumption that post combustion occur from leakage air and at high lance positions for the gas that have already left the open system.

A slag weight was calculated as described above in 3.3.2.1. The dataset provided an estimated steel weight but the chromium content of the slag was not given. A material balance of the heat was used to estimate the chromium slag content. All chromium was assumed to be present as Cr2O3.

The interaction with refractory was neglected in the calculations. Also, iron leaving as iron oxide with the gas phase was not considered.

3.3.2.3 Cobalt Calculations

The approach used to calculate the distribution of chromium for the SSAB dataset was employed for cobalt as well. Three trace element analyses of hot metal were provided. The analyses showed cobalt contents of 0.013 %, 0.012 % and 0.012 %. Since the hot metal composition differed between the three analyses it was decided to use the average of the three as input to the BOF. A trace element analysis of scrap was also provided showing cobalt content of 0.016 %. From this it was assumed that all bought and internal scrap had a cobalt content of 0.016 %.

The calculations could not be performed for a specific heat with cobalt analyses for the hot metal, scrap and crude steel. Instead, the ingoing analyses as described above were used for three heats with trace element analysis of crude steel. The heats were chosen to represent a span in reported crude steel carbon content.

(30)

23

The output of cobalt to the gas cleaning system was estimated from trace element analyses of the fine and coarse sludge. It was assumed that 6 kg/t CS dry coarse sludge and 13 kg/t CS dry fine sludge were formed in the process. Cobalt reporting to the slag phase was estimated from BOF slag analyses and calculated slag weights.

4 Results and Discussion

The results are presented and discussed in this section of the report.

4.1 Results of the Blast Furnace Calculations

The results for the blast furnace calculations are presented and discussed below. The sections are divided in the same way as the methods describing the calculations are. 4.1.1 Hearth Equilibrium Calculations

The results for the 2012 dataset are presented in Table 2 through Table 4 below. From the standpoint that this thesis focuses on trace elements it can be argued that only Table 4 is of interest. However, it seems sensible to evaluate an overall performance of the calculations since faults in the prediction of major components may affect the distribution of the trace elements.

Table 2 presents the comparison of calculated hot metal composition and actual composition. It can be seen that the carbon content of the hot metal is overestimated. In the calculations, crystalline graphite was set to represent the dead man instead of amorphous coke which could explain the difference. However, the value is still considered to be satisfactorily close to the data. In the calculation of the activity coefficient function for zinc in hot metal the carbon content of the process data was used. To make sure that the thermodynamic calculations can be utilized without relying on values already achieved in the process, the zinc activity coefficient function was recalculated using the value for the overestimated carbon content. Using the new value in the calculations resulted in a decreased hot metal zinc content of 3 %, from 2.7 to 2.6 ppm.

The calculated values of silicon and manganese were both considered to be acceptable considering they were not the main goal of this thesis. It is unfortunate that the value provided for zinc by the data is below the detection limit. Regardless, it is comforting to see that the calculated value is below this limit as well.

(31)

24

Table 2: Hot metal analysis from calculations and blast furnace data for the 2012 dataset.

Element Calculated Actual

%C 5.11 4.67

%Fe 94.2 94.0

%Mn 0.37 0.34

%Si 0.36 0.43

%Zn 0.0003 <0.0005

Table 3 below presents the slag analysis, B2 basicity and slag rate for both the calculations and the data. Considering the simple approach employed, the calculated contents of the major components are satisfactorily close to the data. The calculated ZnO content is well below the ppm range. The result is realistic due to the high temperature, non-oxidizing conditions and zinc oxide activity coefficient in the slag phase. The blast furnace slag analysis of ZnO is excluded as the value provided in the balance was 82 ppm. Analyses of crushed blast furnace slag from 2009 to 2013 were provided by SSAB Merox. Out of 26 measurements, twelve were below the detection limit of 4 ppm. In addition to this, eleven was above the detection limit and below 20 ppm. No measurement was close to the level of 82 ppm. Furthermore, the slag sampling made 2006 [33], when the zinc rate was almost three times as high as 2012, all gave values below the detection limit of 4 ppm. Therefore, 82 ppm was considered to be unlikely to represent a mean value for the entire year.

Table 3: Slag analysis, B2 basicity and slag rate from calculations and blast furnace data for the 2012 dataset.

Parameter Calculated Actual

%CaO 37.5 33.3 %SiO2 35.0 33.7 %MgO 14.8 15.1 %Al2O3 12.7 12.9 %MnO 0.07 0.37 %ZnO 0.0000 / B2 1.1 1.0 Slag rate kg/tHM 164 166

Table 4 below presents the material balance for zinc based on the thermodynamic calculations and the actual balance from the data for the year of 2012. The zinc content of the hot metal was set to half the detection limit provided by the data. The slag output of zinc in the blast furnace data was decided to be left out as argued before. The material balance was closed in the thermodynamic calculations, i.e. the input equals the output. The material balance from the blast furnace data suggests that zinc

(32)

25

accumulates in the process over the year. This is of no concern since incomplete balances are common when studying zinc.

The results indicate that the prediction of zinc in hot metal is in the right range, i.e. below the detection limit of the analysis. It is noted that cast house dust is not considered in the blast furnace data material balance.

Table 4: Zn balance from calculations and compared material balance.

Stream Calculated (g Zn/tHM) Actual (g Zn/tHM)

Input

Total input 100 100

Output

Hot Metal 2.7 2.5

Slag 0.0 /

Dust & Sludge 97.3 88.7

Total output 100 91.2

4.1.1.1 Dataset from 2006

Only the final zinc balance is provided for this dataset, Table 5. The complete hot metal- and slag composition are found in appendix 3. As explained for earlier the data is from three days, one of which was operated with lowered top pressure. Therefore, there are great variations in the output of zinc through the top. Furthermore, a balance over this short period does not give a good representation of the output through the top due to the behavior of zinc in the furnace. Also, the output of zinc did not exceed the input although the top pressure was lowered considerably. Thus it was assumed that the recirculation rate was not lowered during the three day period. It was therefore decided to exclude the off gas dust and sludge fraction from the balance to be able to use an average over three days instead of one day when comparing the calculated and actual zinc output in the blast furnace bottom.

The calculated zinc output via the hot metal is underestimated but still within a reasonable range of the data. The cast house dust contains vaporized zinc from the tapping. This means that the zinc content in the hot metal and slag phase is not at equilibrium after leaving the furnace. The lower pressure outside the furnace provides a driving force for the volatilization of zinc. Therefore, the prediction of the calculations is worse since the cast house dust should be accounted for in the calculated slag and hot metal output.

The output in the slag as given by the blast furnace data is based on half the detection limit of the analysis. From the arguments regarding ZnO in section 5 below together with the calculations it is concluded that, from a thermodynamic standpoint, the zinc

(33)

26

content in the slag is below ppm range. Higher zinc contents as reported in section 2.1.2.1 are an effect of other mechanism. It was tried to account for this by fixing the activity of ZnO in the slag. This approach generates problems regarding the material balance of zinc in FactSage and was therefore discarded.

Table 5: Comparison between zinc outputs from the 2006 dataset.

Stream Calculated (g Zn/tHM) Actual (g Zn/tHM)

Input

Total input 290 290

Output

Hot Metal 8.4 12.2

Slag 0.0 0.4

Cast house dust / 2.9

4.1.1.2 Assumptions in the Hearth Calculations

The calculations are directly dependent on the assumed recirculation of zinc in the furnace and not knowing the accuracy of this assumption is devastating for the confidence in the results. Both calculations presented above gave results in the range of the data, although one measurement was below the detection limit. To assure that the assumption made in these calculations is valid, more measurements are needed to compare with. It is likely that the circulating load is not constant but rather changes with zinc rate and operation.

Another important assumption having great effect on the calculation results is the amount of gas allowed to be in equilibrium with the melt and slag as this affect the partial pressure of the zinc gas. The effect of this was studied and the results are presented in Figure 8 below. If the mechanism of zinc dissolution in hot metal and slag is assumed to be from the gas phase then there is little influence. This is an effect of that the total zinc content decreases with decreasing gas amount. However, if a fixed amount of condensed ZnO is assumed to enter the hearth, then this input would be the same regardless of the amount of gas in equilibrium. From the figure it is clear that the latter has a significant effect on the calculation results.

(34)

27

Figure 8: Comparison of the effect of blast volume on the zinc content in hot metal with two different assumptions regarding the behavior of zinc.

4.1.1.3 Trends in Zinc Output from Blast Furnace Hearth

The result of increased circulating load of zinc is illustrated in Figure 9 below. The ppm zinc in hot metal increases linearly which is an effect of not using interaction parameter formalism in the description of zinc. The change in output through the slag is insignificant for the zinc balance.

Figure 9: Effect of change in circulating load on output of zinc in hot metal and slag.

The effect of changed temperature on the zinc content in hot metal and slag is depicted in Figure 10 below. The zinc content in the hot metal decreases with increasing temperature which is expected due to the volatility of this element.

0,0 5,0 10,0 15,0 20,0 25,0 30,0 10 20 30 40 50 60 70 80 90 100 pp m Z n i n ho t me ta l

%of total blast in equilibrium with hot metal and slag

Zn dissolved from gas phase Constant amount of ZnO enters the hearth

0,00 0,05 0,10 0,15 0,20 0,25 0,30 0,35 0 10 20 30 40 50 60 0 2 4 6 8 10 12 14 16 18 20 pp m Z n i n Sl ag pp m Z n i n HM Circulating load Zn kg/tHM ppm Zn in HM ppm ZnO in Slag

(35)

28

Figure 10: Effect of change in temperature on output through the tap hole.

Figure 11 below illustrates the effect of changed hearth pressure on the zinc content in hot metal and slag. The increase in dissolved zinc is expected if Le Chatelier’s principle is considered.

Figure 11: Effect of change in hearth pressure on zinc content of hot metal and slag.

From Figure 9 to Figure 11 above it is clear that the combined effect of circulating load, temperature and pressure have large influence on the zinc output through the tap hole. Although there is no data to compare with, the results presented in this section contribute with interesting information whether a change in a parameter pose significant or insignificant changes in zinc output from a thermodynamic standpoint. The effect of changed B2 basicity on zinc content in the slag phase was calculated but it is left out of this report due to the below ppm level content.

0 0,01 0,02 0,03 0,04 0,05 1,5 2,0 2,5 3,0 3,5 1400 1450 1500 1550 1600 pp m Zn O in S la g pp m Z n i n HM Temperature (°C) ppm Zn in HM ppm ZnO in slag 0,005 0,010 0,015 0,020 0,025 1,5 2,0 2,5 3,0 3,5 2 2,5 3 3,5 4 ppm Z nO in S la g ppm Z n in H M

Hearth Pressure (atm)

(36)

29

4.1.2 Thermal Reserve Zone

In the calculations for the thermal reserve zone it was seen that, from a thermodynamic standpoint, zinc does not oxidize and precipitate. For a pressure of 2.75 atm zinc oxide starts to form at a temperature of 738 °C which is located above the thermal reserve zone. The results also show that careful consideration of allowed phases in the shaft needs to be taken as the thermodynamic composition may differ from the real.

4.1.3 Above the Thermal Reserve Zone and below the Throat

The results from the gas cooling study conducted for the region above the thermal reserve zone are presented in Figure 12 below. The reported temperature interval where zinc condensation begins as described in section 2.1.2.1 does not agree with the results achieved in these calculations. The interval could be intended for other conditions regarding top pressure and %ηCO. In the provided data it is found that the mean top gas temperature of 2012 for blast furnace no. 3 at SSAB Luleå was 133 °C with monthly averages up to 150 °C. This means that thermodynamically, zinc does not leave the blast furnace in the top.

Figure 12: Gas cooling study above the thermal reserve zone. The zinc as gaseous zinc and solid zinc oxide is given as a percentage of the total amount of zinc. The units are presented in brackets to avoid confusion with solidus and gaseous notation.

It is well known that zinc exits the blast furnace via the top gas in both the sludge and dust fraction for furnaces operated with or without top pressure. It was therefore attempted to use the calculations presented above as a base when estimating the outgoing zinc. Considering the ideal temperature profile of the blast furnace as

2,00 2,05 2,10 2,15 2,20 2,25 2,30 0,0 10,0 20,0 30,0 40,0 50,0 60,0 70,0 80,0 90,0 100,0 500 550 600 650 700 750 800 850 900 Pre ss ure [a tm] Zn (g ), Zn O (s ), E ta CO [% ] Temperature [°C]

(37)

30

depicted by Figure 2 it can be seen that the gas temperature decrease rapidly just below the throat. From the figure it is estimated that the gas has a temperature of 650 °C at a level of 2.5 m below the throat. This temperature is chosen as a cut-off temperature where:

1. The already condensed zinc oxide travels down the furnace with the burden, and

2. The zinc gas condensing when ascending to the top travels with the gas phase out of the furnace.

By doing this, 91.7 % of the circulating load will travel down the furnace, continuing the zinc cycle. 8.3 % or 83 g/tHM would leave through the top. This is not far from the value of 88.7 g/tHM which was presented in Table 4.

The effect of changed top pressure on zinc condensation during the gas cooling is illustrated in Figure 13 below. The figure illustrates how lower top pressure is less favorable for zinc oxidation and condensation. In the full scale trials conducted in [33], a decrease in top over pressure from 95 kPa to 65 kPa resulted in an increased average zinc output of 100 g/tHM in the blast furnace top. In that case it corresponded to a 91 % increase. Choosing 650 °C as a cut-off temperature for top pressures of 2 atm and 1.5 atm in the figure below it was calculated that the lowered top pressure increased the zinc output by 31 %. The major part of the increased output when lowering the top pressure is consequently explained for by other factors than thermodynamics. Lowering the top pressure increases the gas flow through the furnace which results in an increased amount of dust and sludge. A change in either blast rate or top pressure can therefore be interpreted as a change of at which distance from the throat the condensed zinc oxide travels with the gas phase out of the furnace. One way to combine the thermodynamics with furnace parameters is to develop a way to describe the cut-off temperature as a function of e.g. blast rate and top pressure. The data available for comparison is scarce (two operational points to compare with) where after such a function could not be designed in this thesis.

References

Related documents

Ä ven en viss konservatism vid val av metod och fasthållande vid ett m indre antal beprövade metoder kom m er enligt min uppfattning att i längden läm na

At the age of 28 days, the mortars were tested in a steady-state diffusion cell, to measure the chioride ion diffusion. The diffusion coefficient was estimated from this test, with

Using the estimated coal mass flow as the measurement output and the valve as actuator a mathematical model of the injection process is identified from process data.. The obtained

There are two ways to measure the coal flow in an injection line, by using a flow meter, e.g., a Coriolis flow meter, or by extracting flow information from a video image of the

Ett antagande kan vara att patientsäkerheten samt den kan brista när sjuksköterskor upplever arbetssituationen som stressande och sjuksköterskan kan då riskera att

Chemical reactivity during combustion and gasification is influenced by the particle size of the material but due to the other differences between the materials (carbon

 What are the differences between the variables Dielctricity, Electrical Conductivity and temperature of Air Cooled Blast Furnace Slag, which was used in Luleå and

In the case of a peripheral pressure distribution, the distortion created a free layer high in the centre and low along the periphery. However, here the particles at the wall