HIsmelt, Adapted Technology for Ti/V-Magnetite Jacques Pilote Manager, Technical Projects HIsmelt Corporation (Pty Ltd) part of Rio Tinto (Plc & Ltd) Abstract The HIsmelt process represents one main alternative to the blast furnace. At the heart of the process is splash-driven heat transfer between the upper (high oxygen potential) combustion zone and the lower (low oxygen potential) smelting zone. This unique high intensity smelting configuration, originally developed for more conventional ore bodies (i.e. haematite and magnetite), is particularly interesting for non-conventional ore types such as titano-magnetite. Several drawbacks are encountered with the current blast furnace process for treating Ti/V magnetite. These include deterioration of the sinter properties, difficult handling of slag and sticky metal due to reduced Ti species formed in the highly reducing atmosphere of the BF process. These can be overcome by HIsmelt without using dilution of the ore with low titania feed or heavy fluxing strategies as used by most blast furnace operators. The current paper presents an initial investigation regarding this version of HIsmelt. The objective is to couple the low capex advantage with low operating costs for this very important resource. Key words: Titano-magnetite, Iron sands, Alternative Iron making, Vanadium dross, Titanium pigment feed Introduction Until recently, to keep growing with minimum risk, integrated steel mills have successfully built larger production blast furnaces. Whilst this improves overall cost structure, it also brings more demanding specifications for raw materials. These large production units are also very difficult to turn down. In a stable economy where production is in balance with demand this is not a major issue. However the recent mineral boom exposed how far resource pricing can go - the GFC that followed also exposed the value of flexibility in turn-down and smaller operating units. In both cycles, someone who had potentially invested in new technologies would have seen significant benefits. The blast furnace is a very mature and efficient technology. Over the past decades the best iron and steelmakers had also learned how to cope with higher alumina and higher phosphorus ores. To some degree this has diminished one of the original drivers for classical HIsmelt. However, in a titano-magnetite context the blast furnace is still poorly adapted due to its highly reducing condition - this becomes its main drawback for smelting Ti/V magnetite feeds. Very low oxygen potential at the metal/slag interface leads to difficult slag/metal handling issues. Over time very motivated ironmakers have learned to cope with this by (i) adjusting their practice to target oxygen potential where Ti(C,N) is sufficiently destabilised (ii) diluting the sinter head feed with low titania iron ore (e.g. Australian iron ore).
The HIsmelt process has been developed to provide an alternative to the blast furnace. The key metallurgical point of differentiation between HIsmelt and the blast furnace is that HIsmelt operates with significantly higher oxygen potential in the slag. This is the underlying reason for HIsmelt s ability to reject phosphorous in slag, and this can also be applied to titania. The result is a metallurgically cleaner process, with avoidance of reduced titania species and more highly oxidized titania in slag. Blast Furnace Limitations for Ti Bearing Ore The smelting of an iron ore containing significant amounts of titania (>5%) is challenging. When used at high proportion (>30%) in the sinter feed, there is enough titania (and its associated high level of alumina) to disrupt sinter properties. Sinter plant productivity drops significantly and sinter quality declines (leading to high decrepitation rates which impede shaft permeability and furnace stability). In addition, at this level of titania, the slag is strongly reduced and becomes viscous as a result of titania being reduced to magnéli phases and Ti(C,N). In small quantities this can be beneficial to the overall life of the hearth in a regular blast furnace. However, in a BF with high titania feed, it can easily reach an intolerable level. For these reasons, even the most skilled blast furnace operators tend to use both dilution and heavy fluxing strategies [1]. High Alumina and Titania are detrimental to the Sinter Quality Cohesive zone requires a low viscosity slag (~2 poises) Highly reducing condition in the coke bed leads to reduced Ti species that increase the slag viscosity and compromises the hearth drainage and metal/slag separation in the trough Figure 1 Main Blast Furnace drawbacks HIsmelt Background The HIsmelt process was conceived in the early 1980 s and progressed through two levels of pilot plant (10,000 and 100,000 t/a respectively) to its first commercial 0.8 Mt/a installation
in Kwinana, Western Australia [2,3]. It originally evolved from a modified K-OBM/KMS modified into a smelting reactor. This evolution legacy is still present in the current configuration in the sense that it offers an optimised reducing condition originally exploited for phosphorus removal (but also useful for smelting Ti/V bearing ores). At the core of HIsmelt is the Smelt Reduction Vessel (SRV) which is illustrated in Figure 1. It consists of a water-cooled upper shell and a refractory hearth. The process uses highvelocity injection of coal and ore into the melt via downwardly-angled water-cooled injection lances. Injected coal, after heating and devolatilisation, dissolves to maintain around 4% carbon in metal. Injected ore is then brought into contact with this carbon-rich metal, and smelting occurs. The lower part of the SRV is maintained at low oxygen potential to allow this reaction to occur, and reduction kinetics balance out at around 5-6% FeO in slag. Heat supply to maintain the necessary thermal balance comes from combustion of bath gas (mainly CO) in the upper part of the vessel. Oxygen-enriched hot blast (typically 35% total oxygen at 1200 C) is introduced via a top lance, and combustion occurs in the relatively oxidizing region in this upper section. The resulting process offgas is typically has a postcombustion degree of 50-60%. The key to the HIsmelt process is achieving heat transfer between the upper (oxidizing) region and the lower (reducing) region in such a way that the oxygen potential gradient is maintained. This is achieved via large amounts of liquid slag splash moving between regions, carrying heat with them as they go. A certain percentage of this heat goes to water panels and lances, and the balance is used for smelting. Offgas Hot Blast Coal Water-Cooled Panels Ore Slag Metal Refractory- Lined Hearth Figure 2 HIsmelt SRV Slag is tapped periodically via a water-cooled notch, while hot metal is tapped continuously via a forehearth. The latter is considered a key safety feature of the technology, since it is the primary means by which metal level is controlled to ensure there is always a suitable safety margin between it and the water-cooled lances.
What HIsmelt Offers for Ti Bearing Ore The HIsmelt process configuration has a number of unique features that can be briefly describe as following: 1. The method of solids injection, using high-speed lances, means that capture efficiency in the melt is very high and even ultra-fines can be used directly. There is no limitation imposed by the properties of the sinter. 2. The natural FeO level in the slag (5-6%), in conjunction with metal carbon at around 4%, creates compromise conditions for avoiding the TiO x to corrupt the metal to slag handling and still offers a reasonable recovery of the vanadium units. 3. The process is also naturally better with more viscous slag. Petrographic examination of the classic HIsmelt slag revealed that the optimum level of solid in the slag is around 20%. At this level, the slag still behaves as a liquid good for slag tapping but exhibits enough viscosity to efficiently coat the water cooled panels. This last aspect is necessary to minimise the heat loses, to maintain a strong central fountain and to protect the refractory by the presence of a freeze lining. 4. The slag and the metal are tapped separately by batch through a water cooled slag notch and continuously via a metal forehearth respectively. This allows, by taking advantage of the natural separation occurring in-situ in the reactor, to produce a clean continuous flow of hot metal. The residual metal contained in the slag in form of prills settles in the slag pit and is recuperated as beech iron. These features can be exploited to access lower-grade feed materials which would be difficult (or impossible) to use in a blast furnace. The next section of the papers describes in more detail the theoretical considerations supporting this. Slag Chemistry Operating Window Classical HIsmelt slag chemistry has been optimised to operate within the range where the main component that firstly precipitates is melilite (see figure 3) [4]. This allows the process to operate slightly sub-liquidus without a drastic increase in the slag solid fraction. HIsmelt slag can be compared to a classical blast furnace as longer (more viscous) slag. With regard to slag viscosity prediction, the progression of solid fraction over a range of temperature is particularly important. The viscosity model used in Figure 4 is a modified version of the FactSage viscosity package (accessed under restricted release to its main sponsors [5]). This model is valid for liquid slag and has been modified to include the viscosity increase associated with solids in sub-liquidus slag melt [6]. The modified viscosity equation is: where c is the solid fraction. µ Slurry = µ FactSage of liquid * (1+ 2.5 c + 9.15 c2)
For the titania version of HIsmelt operation, the slag has been redesigned for additional TiO x components. In comparison of a blast furnace operator approaching this task, HIsmelt has two main advantages: - The process has demonstrated its resilience with regard to high viscosity slag (acceptable viscosity range for HIsmelt is in the range of the 2.5 to 5 poise when the upper limit for a Blast Furnace is around 2.5 poise) - The TiOx will be present mainly in the form of TiO 2 due to the high oxygen potential present in the slag. The oxygen potential in HIsmelt slag is estimated at 17x10-9 Pa [7], approximately 3 times higher than Panzhihua Blast Furnace measured by EMF in laboratory furnace. At this level all reduced titanium species (e.g. Ti(C.N) in slag and metal, Slag Ti 3+ and Ti 2+ and metal Ti) are unstable and reports in large part the slag as TiO 2. Good indicators of this trend in HIsmelt process are: low silicon and phosphorus level in hot metal and the low sulphur partition ratio. However, HIsmelt has a very unique feature that mitigates this drawback, the large part of the sulphur (75%) reports to the off-gas which scrubbed by a Flue Gas Desulphuring process. Furthermore, there is room to accept some level of Ti 2 O 3 that may be present if the SRV operated with higher levels of TiO x. This can be valuable in the case of a Ti/V ore body which has an associated ilmenite stream. By dosing the SRV with a higher titania level (ie blending some ilmenite into the feed), the slag can potentially be turned into a pigment feed material. This could significantly improve the overall economics by valorising the slag as a co-product. Some exploratory tests have been carried out with TiO 2 levels around 60%, and these indicate a reasonable operating window in term of slag liquidus control. This suggests it may be quite a realistic option (see figure 5) [8]. HIsmelt also provides some practical advantages for coping with more viscous slag (compared to a blast furnace). In the SRV, slag is tapped separately from the hot metal. This means fast metal-slag phase separation in a trough (as per the BF) is not needed. Clean metal is tapped continuously through the forehearth where relatively low carbon levels avoid the formation of TiC/TiN, but still offer enough reducing potential to achieve reasonable vanadium recovery (see below). Vanadium Recovery in HIsmelt The balance between the oxidising condition of the slag and the reducing condition of metallurgical processes can be approximated by the comparisons of their respective Slag total Fe / metal Carbon ratios. By comparing a wide range of TiO 2 bearing ore smelting processes (EAF ilmenite smelting and Ti/V magnetite feeds in the EAF as well as in BF) it is possible to derive a relationship between the vanadium partition and [Fe t ]/(Cmetal) - see figure 6. Considering the relative position of HIsmelt on this plot, it is estimated to obtain a vanadium partition of the one achieved in the EAF. Considering that HIsmelt doesn t require dilution of the Ti/V magnetite ore burden or heavy fluxing to achieve acceptable slag chemistry, this lower dilution of the vanadium units can also be translated into high vanadium recovery. Some research work also points out that, under oxidising conditions, there are some vanadium units present in slag as VC [9] - the oxidising condition would tend to destabilise
these carbides. In such a case, the regression curve presented below would represent a conservative figure rather than an average prediction. Figure 3 Classical HIsmelt Slag Tertiary System Comparison of Slag Viscosity Viscosity (poise) 7.00 6.00 5.00 4.00 3.00 2.00 1.00 Still room to move Classical HIsmelt 30%Ti/V Magnetite in Blast Furnace Ti/V Magnetite HIsmelt Classical Blast Furnace 0.00 1300 1350 1400 1450 1500 1550 1600 Temperature (Deg C) Figure 4 Slag Viscosity Prediction for different slag chemistries
Option for TiO2 Pigment Feed 100% Titanomagnetite Titanomagnetite + Ilmenite Operating Valley for HIsmelt Figure 5 Slag liquidus measurements on high TiO 2 slag 100% Partition ratio (slag/metal) of different TiO 2 bearing process EAF on Ilmenite 90% 80% 70% EAF on Titanomagnetite BF Process [Vs]/(Vm) Index 60% 50% 40% 30% HIsmelt 20% 10% 0% 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Slag Fet/Metal C Figure 6 Vanadium partition index for different TiO 2 bearing ore processes Conclusions In addition to the classical cost drivers for HIsmelt (reduced exposure to future coking coal prices, ability to process high phosphorous/alumina feeds and lower capital cost), HIsmelt is also proposing suitability for new type of ore which is currently considered a second choice for the industry. Under this scenario Ti/V magnetite complex ore bodies become
more attractive not only for their iron content, but also vanadium and potentially pigmentgrade slag co-products. In order to develop this option, HIsmelt will conduct small-scale (around 1 t/h) operating campaigns with the proposed slag in 2011. Further development will follow at larger scale. The value proposition is potentially large. Although this version of HIsmelt has not been fully proven, the size of the prize may justify a certain degree of risk-taking. The lowestrisk commercial implementation of HIsmelt (today) involves conventional magnetite cold feed (as dried concentrate, with no ore preheater). If it is possible to configure a commercial project with flexibility to feed both conventional magnetite and Ti/V-magnetite, then the risk profile of such an investment may allow such a project to move ahead immediately. In the absence of such an early mover, HIsmelt intends to develop the Ti/V option in stages, moving from small to larger scales over time whilst further deepening its fundamental understanding via supplementary investigations. HIsmelt is a young technology, full of promise. It hasn t revealed its full potential yet! List of References 1. Du He-gui, Theory of Smelting of Magnetite Bearing Vanadium and Titanium in Blast Furnace, 1996 2. P. Bates and A. Coad, HIsmelt, The Future in Ironmaking Technology, 4th European Coke & Ironmaking Conference, Vol. 2, June 2000, pp. 597-602. 3. R J Dry, C P Bates and D P Price, HIsmelt - the future in direct ironmaking, Proc 58th Ironmaking Conference, Chicago, 21-24 March 1999, p361 4. C.W. Bale, A.D. Pelton & al, Ecole Polytechnique de Montreal, FactSage 6.1 database 2008 5. N Wan-Yi Kim, Arthur D. Pelton and Sergei A. Decterov, Ecole Polytechnique, Montréal, Modeling Viscosity of Silicate Melts Containing Titanium Oxides draft paper supporting FactSage viscosity 6.0 model 6. M. S. Oh, E. F. de Paz, D. D. Brooker, J. J. Brady, and T. R. Decker Texaco, Inc, Effect of crystalline phase on coal slag viscosity 7. E. Heikinheimo, Solid State Ionics 3/4 North Holland Publishing (1981) 541-545 8. B. Zhao, P. C. Hayes and E. Jak University of Queensland, Experimental determination of phase diagrams for HIsmelt high titania slag, July 2010 9. Wang, Ming-yu & al Separation of Iron Droplets from Titania Bearing Slag, Journal of Iron and Steel Research, International, 2008, 15(1);45-48