CHAPTER 2 ILMENITE SMELTING PROCESS OVERVIEW
The purpose of this chapter is to introduce the reader to various aspects regarding the ilmenite smelting process. It also combines information from a number of literature resources to provide a consolidated body of data and information on which the work of this study is based.
Although more than one type of furnace design is in industrial use for ilmenite smelting, the current work focuses on a DC furnace like those operated by Namakwa Sands and Ticor South Africa. A schematic of such a furnace is shown in Figure 8.
Because of the secrecy surrounding ilmenite smelting technology (Rosenqvist, 1992) and the resulting lack of published information, the descriptions provided here are for the most part qualitative. Even though this is the case, these descriptions should serve well to introduce a reader that is unfamiliar with the process.
The process is operated semi-continuously. Ilmenite, reductant and electrical energy are discharged into the furnace as continuously as possible. The operation is only stopped for adding electrodes, planned and unplanned maintenance, and for process-related incidents like severe slag foaming. Off-gas is extracted from the furnace continuously, while slag and metal are tapped separately in batches.
Ilmenite and reductant enter the furnace through a single hollow graphite electrode that extends down into the furnace through the centre of the furnace roof. Since the electrode is consumed as it delivers energy, it acts as an additional carbon source. Energy is discharged into the furnace via an open arc between the tip of the electrode and the bath. The only other inputs into the process include air drawn in during periods of negative furnace pressure, and water when leaks are present in the furnace roof, off-gas duct or other water-cooled elements.
Ilmenite is expected to melt quickly as it travels through the arc and into the slag bath. This is due to its relatively low melting point of 1397 °C (FactSage 5.2) and its small particle size of between 100 and 1000 µm. As reductant particles are heated up in the arc and slag bath, these particles devolatilise and possibly break up due to thermal shock. Reduction reactions are expected to take place at the interface between liquid slag and reductant particles. The products of these reactions include gas (primarily CO), slag (enriched in TiO2 and Ti2O3) and iron metal. Some reduction reactions also occur at the interface between the slag and metal bath because of the high carbon content (around 2%) of the metal bath. The products of these reactions are the same as for reduction with reductant particles.
Because of the density difference between metal and slag (metal being denser that slag), metal collects at the bottom of the furnace as a metal bath and slag on top of it as a slag bath. Small metal droplets are present in the slag bath due to the reduction reactions taking place there, but these droplets continuously find their way down to the metal bath or are reabsorbed into the slag. Gas bubbles forming in the slag bath due to reduction reactions rise up through the slag bath and escape into the furnace freeboard.
Metal is extracted from the furnace through a metal tap hole and slag through a slag tap hole. The slag tap hole is situated at a different angular position than the metal tap hole, and higher in the furnace wall than the metal tap hole. This is because the slag bath is floating on top of the metal bath. Off-gas, fumes and dust that continuously rise up into the furnace freeboard are extracted by the off-gas system through a duct installed in the furnace roof.
The off-gas system is operated in such a way that a slight positive pressure is maintained in the furnace freeboard. This is done to prevent air from being drawn into the furnace. Air consumes carbon and oxidises slag, resulting in reduced reductant efficiencies and electrode wear.
The operational personnel and control systems of an ilmenite smelting furnace have the following two primary control objectives:
- Maintain a safe freeze lining thickness to ensure protection of the furnace refractory lining against chemical attack by liquid slag.
- Produce slag with a titanium dioxide content that is within the provided chemical composition specification.
The controlled variables from these objectives are:
- Furnace sidewall temperature
Sidewall temperature is used as an indirect indication of freeze lining thickness because the state of the freeze lining cannot be measured directly.
- Slag TiO2 content Chemical analyses of tapped slag are used as a measure of this controlled variable.
The manipulated variables used to effect control of the controlled variables are:
- Specific energy input (kWh/ton ilmenite)
- Specific reductant input (ton reductant/ton ilmenite)
Since both manipulated variables have an influence on both controlled variables, the control problem is of the multiple-input-multiple-output (MIMO) type. This together with the dead time associated with measurement of both controlled variables, makes this an interesting control problem that is difficult to solve. It has been reported that the two manipulated variables cannot be changed independently (Pistorius, 1999). This is due to a balance between the enthalpy of the reduction reaction and the enthalpy required to keep the slag at or just above its liquidus temperature. This relationship between the two manipulated variables can be used to simplify the control problem to some degree.
The furnace vessel is shown in Figure 8. It is used to contain the metallurgical process of ilmenite smelting.
The vessel is divided into the following four parts for discussion here:
- Tap holes
The hearth of a DC furnace has two roles. The first is to act as the floor of the furnace, containing the metal bath present directly above. The second is to provide the anode connection of the DC electrical circuit.
To contain the hot liquid metal, multiple layers of refractory brick are contained in a steel shell. The bricks are precision manufactured and installed to ensure that no metal will penetrate the lining and burn through after the lining has been heated up and the bricks have completed thermal expansion.
The hearth bricks are usually chosen to have high thermal conductivity, similar to the sidewalls. Air cooling or water cooling is used to extract heat from the outer surface of the hearth and cool it down. Thermocouples are installed in the hearth to generate temperature signals for monitoring the hearth’s condition.
The anode connection of the DC electrical circuit can be integrated into the hearth through a number of anode designs. Examples include fins or concentric metal rings built into the anode, metal pins, metal billets and a conductive hearth. The last design is the only one where the contact between the metal bath and the anode is not metal, but rather through electrically conductive refractory brick.
The sidewalls of an ilmenite-smelting furnace also have a dual purpose. Firstly, similar to the hearth, it must contain the metal and slag baths and the furnace freeboard. Secondly, the sidewalls must extract heat from the slag bath at a high enough rate to establish and maintain a freeze lining of solidified slag on the wall.
The sidewalls are constructed from refractory brick, a steel shell and often a layer of ramming material that is inserted between the bricks and shell to prevent air gaps and ensure good thermal contact. This is especially important on the level of the slag bath where heat extraction is critical.
To further ensure adequate heat extraction on the slag bath level, refractory brick with high conductivity and high density is chosen and water cooling is applied on the outside of the steel shell.
Thermocouples are installed at various vertical, angular and radial positions in the sidewalls. Signals from these instruments are used to monitor the condition of the sidewall refractory lining, and as an indirect measure of freeze lining thickness.
The roof of the furnace acts as a lid, closing the furnace so that no gas, fumes or dust escapes other than through the off-gas duct. Because the slag bath surface is fluid and at temperatures of between 1650 and 1750 °C, heat is radiated to the furnace roof at high rates. For this reason the roof is water cooled.
One or more inspection ports are usually installed into the roof to enable operational personnel to do visual inspections of the furnace interior.
Tap holes are installed into the furnace sidewalls. One or two metal tap holes are installed on the level of the metal bath, and another one or two slag tap holes are installed at a higher elevation on the level of the slag bath. The tap holes assemblies usually include water cooling aimed at preventing overheating and excessive wear of the refractory lining in the region of the tap holes.
Opening a tap hole is usually done by drilling into the wall up to a specific depth and then using an oxygen lance to burn through to the bath (Noda, 1965). It is closed with a clay plug that is pushed into the hole with a clay gun.
Ilmenite is the primary input material into the furnace since the purpose of the process is to upgrade ilmenite to a product of higher TiO2 content. Both DC ilmenite-smelting furnace operations (Namakwa Sands and Ticor South Africa) utilise sand-type ilmenite as feedstock. The particle size of this ilmenite is usually less than 1000 µm.
The major phase in the ilmenite feed is, of course, ilmenite (nominally FeTiO3). Depending on the ore body, some FeO could have been weathered out to leave some rutile (nominally TiO2) or pseudorutile (Fe2Ti3O9) (Nell, 1999). It is also possible that the ilmenite feed could have been roasted to increase the magnetic susceptibility of ilmenite grains over that of unwanted impurities such as chromium-bearing spinel (nominally FeCr2O4) and garnet ((Fe2+, Mg, Mn, Ca)3(Al, Fe3+)2Si3O12) (Nell, 1999). Such roasting is usually oxidising and could result in some of the ferrous iron being converted to ferric iron. The end result is some Fe2O3 in an M2O3 solid solution with FeTiO3 together with TiO2 as a separate phase.
Together with the titanium and iron oxides, numerous impurity chemical species also occur. These include CaO, MgO, MnO, ZrO2, SiO2, Al2O3, Cr2O3, P2O5, V2O5, Nb2O5 and possibly others. Some of these species (MgO, MnO) can occur as part of the ilmenite solid solution, while others (SiO2 and Cr2O3) report as separate phases like spinel and garnet as mentioned above. The levels of the various impurities depend on the ore body.
All of the above effects often result in natural ilmenite containing less TiO2 than the 52.6% by mass of stoichiometric pure ilmenite (FeTiO3).
The reductant used in ilmenite smelting is firstly chosen to have low ash content because virtually all of the ash reports in the slag phase as impurities. Secondly, low volatile matter content is preferred. This is to ensure a lower rate of gas evolution from the reductant that can contribute to slag foaming without significantly contributing to reduction. Anthracite and charred coal are therefore mostly used as reductant.
The particle size of the reductant is chosen relatively fine, less than 10 mm. A fine particle size is beneficial for reaction rates in the furnace due to the increased surface area. Too-fine particles are not favoured because these can blow out of the furnace with the off-gas and reduce reductant efficiency
The graphite electrodes used in DC ilmenite-smelting furnaces are pre-baked, consisting of more than 99% graphite. The amounts of moisture, ash and volatile matter in these electrodes are negligible. They are manufactured with a hole in the centre through which feed material enters into the furnace.
Various refractory materials are used in different locations in the furnace lining. All of these will not be discussed here. The material of major importance to this work is the refractory brick used in the sidewalls on the level of the slag bath. These bricks are required to have high thermal conductivity so that heat can be extracted through them to maintain a safe freeze lining thickness. High density, high-MgO material is usually used.
The reason for discussing refractory material under the heading of process inputs is the fact that some of this material is inevitably released into the slag bath through wear of the lining. This is significant because very strict MgO content specifications are usually in place for the slag product destined for pigment production through the chloride process. The reasons behind these strict specifications are explained in paragraph 1.1.7 (page 7).
Leaks in the furnace roof and off-gas duct can result in water entering the furnace. This has a dual effect on the process. Firstly, the water liquid absorbs energy from the process and is converted to water vapour. The energy associated with this effect reduces the energy efficiency of the process. Secondly, the addition of water vapour increases the oxygen partial pressure in the furnace. This adversely affects reductant efficiency. Both these effects are negligible if only small amounts of water are considered, but it could become significant for increased volumes.
Other, more dramatic influences of water include the risk of explosions in the furnace when the conversion from liquid to gas somehow occurs in a confined space; and hydration of the MgO bricks that can severely damage the refractory lining and reduce its life.
During normal operation, air is drawn into the furnace through gaps between furnace roof segments. It is attempted to control the furnace atmosphere at a slight positive pressure, but conditions in the furnace still result in periods of negative pressure.
The oxygen in air can strongly increase the oxygen partial pressure in the furnace. This can be detrimental to reductant efficiency. In terms of energy, the net effect of air entering the furnace should be positive. The heat absorbed by the inert N2 is less than the heat released by O2 reacting with CO to form CO2. This is of course only true if air is not added in excess to what is required for the combustion of CO. This is most likely the case during normal furnace operation.
Large volumes of air are drawn into the furnace when it is placed under negative pressure while an electrode is added or one of the inspection hatches is opened. Under these conditions, air is likely to be present in excess, and it will contribute to cooling the furnace down, consuming some unreacted reductant and possibly oxidising some of the species in the slag.
Energy can enter the furnace in two ways. Since the furnace is equipped with a DC power supply, discharge of energy through the electric arc is the primary means of energy input. In addition to this, both DC ilmenite-smelting furnace operations (Namakwa Sands and Ticor South Africa) equipped their furnaces with ilmenite pre-heaters. These pre-heaters utilise the chemical energy in the CO rich furnace off-gas to heat ilmenite before discharging it into the furnace.
The furnace is operated with a slag bath temperature of around 1700 °C (Geldenhuis and Pistorius, 1999). Slag is therefore tapped from the furnace around this temperature. The target total TiO2 content (all Ti reported as TiO2) of the slag is around 85% (Kahn, 1984).
Refer to paragraph 2.6 for a discussion of the properties of titania slag.
Heat is extracted at a significant rate from the metal bath through the hearth and sidewalls. For this reason the temperature of the metal bath can be up to 150 °C lower than that of the slag bath (Pistorius, 1999). The carbon content of the metal is around 2% (Pistorius, 1999). It also contains other elements such as Si, Mn, S and P. The levels of such elements in the metal tend to increase as the level of reduction of the slag is increased.
The major species in the off-gas include CO (estimated at between 80 and 90%) as a product from the reduction reactions, and H2 (estimated at 5 to 15%) from volatiles in the reductant. Some CO2, H2O, N2 and SO2 are also found. Because of the high CO and H2 content, this gas is rich in chemical energy. For this reason it is recycled for purposes such as ilmenite pre-heating.
Dust and Fumes
It is likely that thermal shock is experienced by materials as they enter the zone of the electric arc and the bath just beneath it. This can cause reductant and ilmenite particles to break up, producing fine dust particles that are carried into the furnace atmosphere. This is a likely reason for the thick dust cloud present in the furnace freeboard during normal operation. In addition to this, species like Mn and SiO are fumed off at the operating temperatures of the furnace. The dust extracted from the furnace by the off-gas system is enriched in these species.
Energy leaves the furnace via a number of routes. Liquid slag (at around 1700 °C), liquid metal (at around 1600 °C), and off-gas and dust (at around 1700 °C) carry out large quantities of sensible heat. In addition to this heat is lost through the hearth, sidewalls and roof.
PROPERTIES OF TITANIA SLAG
Because of the specific importance of titania slag to this project, the relevant properties of this material are discussed below in more detail.
Valuable work on the thermodynamic properties of the Fe-Ti-O system has been done in recent time. Firstly thermodynamic data from numerous authors were critically assessed and optimised for a number of TiO2-containing binary systems (Eriksson and Pelton, 1993). The result of this study was, for each binary system, a single set of modified quasichemical model equations describing the Gibbs energy of the liquid slag phase as a function of composition and temperature, and equations for the Gibbs energy of each compound as a function of temperature. The two systems in that work that are of interest to this study are the FeO-TiO2 and Ti2O3-TiO2 binaries.
The 1993 work was followed up with some experimental work on the stability of pseudobrookite solid solutions in the FeTi2O5-Ti3O5 range, combined with critical assessment of more literature data (Eriksson et al., 1996). All this data was combined and optimised to yield a comprehensive description of the thermodynamic properties of the FeTiO3-TiO2-Ti2O3-Fe system. This is the system of primary importance to the current study. The data from these studies have been included into the FactSage 5.2 software system (Bale et al., 2002). This software system and the data in its databases were used for much of the work in this study.
In 1999, Pesl and Eriç did measurements in the TiO2-FeO-Ti2O3 ternary system. These measurements have not been incorporated into the FactSage data. For this reason only comparisons between the Pesl and Eriç data and the FactSage calculated values are presented here.
TABLE OF CONTENTS
LIST OF FIGURES
LIST OF TABLES
CHAPTER 1 INTRODUCTION
1.1 INDUSTRY BACKGROUND
1.1.2 Hard Rock Mineral Processing
1.1.3 Mineral Sands Processing
1.1.4 Ilmenite Smelting
1.1.5 Synthetic Rutile Production
1.1.6 Sulphate Process Pigment Production
1.1.7 Chloride Process Pigment Production
1.1.8 Titanium Metal Production
1.1.9 Importance of Ilmenite Smelting
1.2 OVERVIEW OF CURRENT WORK
1.2.2 Impetus for Study
1.3 ORGANISATION OF CURRENT TEXT
CHAPTER 2 ILMENITE SMELTING PROCESS OVERVIEW
2.1 OPERATIONAL OVERVIEW
2.2 FURNACE CONTROL
2.3 FURNACE VESSEL
2.4 PROCESS INPUTS
2.5 PROCESS OUTPUTS
2.6 PROPERTIES OF TITANIA SLAG
CHAPTER 3 A 1D MODEL OF THE FREEZE LINING AND FURNACE WALL
3.2 PROBLEM DEFINITION
3.3 SYSTEM DESCRIPTION
3.4 KEY PHENOMENA
3.5 APPROACH AND MODEL COMPLEXITY
3.6 MODEL FORMULATION
3.7 MODEL SOLUTION
3.8 MODEL VALIDATION
3.9 COMPARISON WITH ACTUAL DATA
CHAPTER 4 A 1D MODEL OF SOLIDIFICATION ON THE SLAG BATH SURFACE
4.2 PROBLEM DEFINITION
4.3 SYSTEM DESCRIPTION
4.4 KEY PHENOMENA
4.5 APPROACH AND MODEL COMPLEXITY
4.6 MODEL FORMULATION
4.7 MODEL SOLUTION
4.8 MODEL VALIDATION
4.9 COMPARISON WITH ACTUAL DATA
CHAPTER 5 A DYNAMIC MODEL OF THE ENTIRE ILMENITE-SMELTING FURNACE PROCESS
5.2 PROBLEM DEFINITION
5.3 SYSTEM DESCRIPTION
5.4 KEY PHENOMENA
5.5 APPROACH AND MODEL COMPLEXITY
5.6 MODEL FORMULATION
5.7 MODEL SOLUTION
5.8 MODEL VALIDATION
5.9 COMPARISON WITH ACTUAL DATA
CHAPTER 6 THE DYNAMIC RESPONSE OF A FREEZE-LINING TO STEP CHANGES IN INPUT HEAT FLOW RATE
6.1 EXPERIMENTAL SETUP
6.3 EXPERIMENTAL RESULTS
CHAPTER 7 THE DYNAMIC RESPONSE OF A FREEZE-LINING TO CHANGES IN SLAG BATH COMPOSITION
7.1 EXPERIMENTAL SETUP
7.3 EXPERIMENTAL RESULTS
CHAPTER 8 THE DYNAMIC PROCESS RESPONSE TO STEP CHANGES IN FURNACE OPERATING PARAMETERS
8.1 EXPERIMENTAL SETUP
8.3 EXPERIMENTAL RESULTS
8.5 THE IMPACT OF SEVERE OPERATIONAL ERRORS
8.6 INDEPENDENT ADJUSTMENT OF REDUCTANT AND ENERGY INPUTS
8.7 APPROPRIATELY COMBINED ADJUSTMENT OF REDUCTANT AND ENERGY INPUTS
CHAPTER 9 THE DYNAMIC PROCESS RESPONSE TO SOLIDIFICATION AND MELTING OF THE FREEZE LINING AND CRUST
9.1 EXPERIMENTAL SETUP
9.3 EXPERIMENTAL RESULTS
CHAPTER 10 COMPOSITIONAL INVARIANCE OF HIGH-TITANIA SLAG
10.1 THE RELATIONSHIP BETWEEN FeO AND Ti2O3
10.2 REACTION EQUILIBRIUM AS A POSSIBLE MECHANISM
10.3 KINETIC EFFECTS AS A POSSIBLE MECHANISM
10.4 EUTECTIC GROOVE PHASE EQUILIBRIUM AS A POSSIBLE MECHANISM
10.5 A NEW MECHANISM PROPOSED
10.6 A REFLECTION ON REDUCTION MECHANISMS
CHAPTER 11 CONCLUSION
11.1 OBJECTIVES AND APPROACH
11.2 CONTRIBUTION OF THIS WORK
11.3 OPPORTUNITIES FOR FUTURE WORK
GET THE COMPLETE PROJECT