In the present invention, attention is focused on major energy savings in direct production of ultra-low carbon iron slab or ingots from low-grade iron ore at the mine site of the future for distribution to steelmakers requiring some input of primary iron to supplement their usage of steel scrap. The way forward is based on melt circulation technology utilizing forced circulation of molten iron in a closed loop by the commercially well-established Ruhrstahl Hereaus (RH) steel vacuum degassing process together with an adaptation of float glass technology for highly efficient heat recovery from the copious reduction slag resulting from elimination of beneficiation.

In a future low carbon economy, a graphite electrode-based electric arc furnace (EAF) is considered as being inappropriate in iron and steelmaking, so the use of electrical conductive heating for melting steel scrap in secondary metallurgy is proposed. Also, to eliminate climate change concerns associated with C0 ₂ emissions, hydrogen must be used as the gaseous reductant for primary iron production from the relatively thin floating oxidic melt layer being transported along on the molten iron carrier medium within the melt circulation loop.

Because beneficiation requires very fine grinding to effect magnetic separation,

considerable energy consumption is inevitable. On the other hand, separation of molten slag from molten iron is not energy intensive. Clearly, the significant issue to be addressed is how can production of the molten reduction slag phase be accomplished in new technology' without a huge thermal energy demand? This must rely on extremely efficient heat recovery from the reduction slag in order to preheat the ore plus flux charged along the whole electrically conductive heated length of an ore incorporation zone in an ironmaking melt circulation loop.

For low-grade taconite or magnetite iron ore, accordingly, the emphasis in the present invention is focused on recovering the massive amount of heat associated with the reduction slag, which up to now has been regarded simply as a means for eliminating gangue components. The inventor has published numerous technical papers on energy recovery from slag. In the present invention, radiative heat transmission is designed to take place immediately the molten slag enters an open channel adjacent to a conveyed layer of ore plus flux undergoing preheating to a temperature of at least 800°C or possibly even higher.

After introducing the molten reduction slag to spread across the width of a molten tin pool, whilst still in the molten state, a shallow layer of molten slag about 7 mm in equilibrium thickness is established and then, after substantial thermal energy transfer radiatively, it is eventually transformed into a thin solid slag slab floating on the tin bath, which continues on its way whilst radiatively cooling in an analogous fashion to plate glass and is eventually induced forward and lifted up out of the molten tin pool by a pinch-roll mechanism and further cooled by inert gas forced convection from about 600°C to around 100°C or even less in what is analogous to the so-called Float Glass Annealing Lehr to cool the glass product carefully down to approaching ambient temperature without induction of residual stress. Both the Float Bath Furnace and the Annealing Lehr are globally commercially well established so their adaptation to primary ironmaking should be relatively straight forward.

It is vitally important to recognize that generic melt circulation technology, with its associated swimming-pool sized reactors, provides ample opportunity for the oxidic melt (slag phase) undergoing hydrogen reduction to become fully saturated with virtually pure iron as it floats on the molten iron carrier medium in the melt circulation loop. Thus, handling the automatic continuous overflow of molten reduced slag (reduction slag) is quite straight forward. The essential ingredient is to have available ceramic-fibre

reinforced iron plate and U-profiled channels to facilitate construction of the slag handling infrastructure necessary to transfer the high temperature slag to the slag continuous casting heat recovery process. The solid iron will not be corroded by the liquid slag as it is already thermodynamically in a state of chemical equilibrium. This ensures reliable commencement of the all important step of radiative heat transfer to a conveyed adjacent layer of solid ore charge plus flux undergoing final preheating before admission to the electrical conductively heated charge incorporation region of the melt circulation loop. The description now focuses on the requirements of an enhanced safety method for producing ultra low carbon primary iron slab continuously without C0 ₂ gas emission directly from iron ore at a mine site with the expectation that the iron slab will be transformed ultimately to a refined steel product at an appropriate location. The methodology is based on generic melt circulation, employing closed loop melt circulation of carbon-free molten iron as the carrier medium, to which is incorporated iron ore to facilitate subsequently pure hydrogen gaseous reduction of a thin layer of oxidic melt comprised initially of liquid wustite (iron-saturated ferrous oxide) as a major chemical component. Electrical heating is used throughout to satisfy the thermal energy demands. The molten iron carrier medium is force circulated around a closed loop, preferably employing a modified gas-lift pumping arrangement, such as used in the (RH) steel vacuum degassing process. The total hearth lengths involved must be relatively long to increase the electrical resistance of the molten iron electrical circuit in order to moderate large electrical currents.

By passing an oxidising gas into molten iron to refine blast furnace metal, Sir Henry Bessemer laid the foundation for modern steelmaking, but the result is reaction intensity far too high for reliable continuous processing. Sub-surface nucleation and growth of carbon monoxide leads to bubbles bursting through the melt surface and creating, in Bessemer's own words "a veritable volcano in a state of active eruption". Hence, emphasis is focused on totally avoiding the presence of carbon in the molten iron.

Adaptation of melt circulation technology is the way forward for future continuous iron and steelmaking. Existing RH vacuum degassing of molten steel is essentially a batch process. Thermal steady-state operation is not reached before the RH unit is lifted out of the ladle of molten steel being refined. This is totally incompatible with the essential needs of truly continuous processing. If the liquid head pumping requirements exceed that delivered in normal RH steel degassing, the solution is to add an extended gas-lift snorkel immediately below and in-line with the normal up-leg of an RH vessel. The individual components for direct iron ore smelting are illustrated in sectional plan views given in Fig.1 and Fig. 2. It is important to realize that the individual "swimming" pool reactors become large static baths of molten iron, if melt circulation ceases, so provision for auxiliary electrical conductive heating throughout the circuit to prevent a freeze-up is mandatory. To effectively achieve this objective, alternating currents (AC) are essential, because of the possible use of thick unmelted steel shells to combat hearth attack by melts with high FeO content. The so-called "skin effect" associated with AC provides the solution. However, control of heat losses to prevent a freeze-up would consume only a minor amount of electric power input from the existing normal supply grid at the mine site. However, there is a reasonable possibility that a fused lining may not in future be needed now that fused hercynite refractory (resistant to molten ferrous oxide attack) is being developed in China.

For electrical conductive heating of the molten iron whilst in continuous operation, attention was focused on the ground breaking progress being made in the primary aluminium industry, where electric currents as high as 600,000 amp DC have already been demonstrated (Rio Tinto Alcan AP Technology™ newsletter issue18, February 2014).

From a health and safety viewpoint, the new hydrogen-based process has been designed so that the whole plant operates at essentially atmospheric pressure. Accordingly, there is no attempt to pre-reduce the iron ore feed prior to its admission to the melt circulation reactor. Also, within the melt circulation loop itself, gaseous reduction only takes place at the base of a hydrogen space above a quiescent floating layer of oxidic melt.

Whilst in operation, hydrogen is constrained from entering the flat refractory roof defining the upper surface of the reducing gas layer by maintaining an appropriate pressure gradient employing a water-vapour protective atmosphere, which acts as an explosion prevention shield, in the structural assembly. This comprises a flat suspended refractory arch and its associated multiple supporting steel joist girders, all rigorously maintained above the steam dew-point. At a temporary shutdown, an inert gas is substituted to take on the role of protective atmosphere. Comprehensive analysis of the process kinetics revealed that at steelmaking temperatures the chemical reaction kinetics in the configuration under discussion has negligible effect on the reduction of a sheet or layer of liquid wustite by hydrogen. The relative effects of gaseous diffusion increase dramatically when the operating temperature is increased above the melting point of wustite, because of a massive increase in the chemical reaction rate constant. Capturing this vital attribute is the key advantage of the new ironmaking technology. The establishment of almost exclusive gaseous diffusion control for iron ore reduction by hydrogen at atmospheric pressure immediately removes the necessity for conducting hydrogen steelmaking at elevated pressures. This clearly has immense safety implications.

Solid magnetite forms liquid iron oxide by the reaction given in Equation 1.

[Fe ₃ 0 ₄ ]soiid + = 4{FeO}ij _qU jd (1 )

Reaction 1 is endothermic and in the present process the heat required is provided electrically by the heavy current flowing in the circulating molten iron carrier medium. The fundamental phenomenon involved, which ensures a successful outcome, is referred to as interfacial turbulence or the Marangoni effect. As a consequence, any liquid ferrous oxide once formed would not stay attached to reacting solid ore surfaces.

Liquid oxide melts containing significant amounts of dissolved iron oxide are notoriously difficult to process, because of extremely aggressive attack on all commercially available refractory materials. Freeze linings or un-melted shells of solid iron are used for melt containment. Once the dissolved oxygen in the molten iron carrier medium exceeds the critical value, where hercynite at unit thermodynamic activity can be formed, freeze linings were considered mandatory. However, the availability of fused hercynite refractory may well become a viable alternative to freeze lining.

Within the melt circulation loop, electrical conductive, heating in regions with massive un- melted shells of highly conductive solid iron in place to facilitate freeze lining should be avoided. Major energy input by conductive heating demands a high circuit electrical resistance, which is obviously destroyed in the presence of such thick steel shells.

Accordingly, conductive heating is necessarily restricted to a zone in which expensive fusion cast alumina with a protective stable layer of hercynite (FeAfeO- can be employed rather than a freeze lining. In effect, this applies throughout the entire ore plus flux incorporation region of the ironmaking melt circulation loop.

Crushed iron ore without removal of ore fines and bauxite flux is distributed uniformly across the full width and of the turbulently open-channel flowing molten iron, whilst being electrical conductively heated in the fusion cast alumina-lined heath of the open channel with to a slag freeze lining limited to a relatively narrow strip precisely where the thin oxidic melt layer poses a threat to the stability of the protective hercynite layer. The location of reduction interface is reliably predictable in this low intensity region of the primary melt circulation loop. This is an overwhelming advantage of generic melt circulation. No other smelting reduction technology can possibly compete in this aspect.

Only one large RH is estimated to be required for melt circulation to produce 2 million tonnes per annum iron product, but for larger tonnages, two or more RH units can be used in parallel. The individual ore particles become coated with a frozen solid iron outer shell. A contiguous layer of pellets is designed not to be established. Thereby sintering together is purposely avoided. The solid iron-shelled particles increase initially in size as the frozen shells increase in thickness and thereafter decease in diameter as they are conveyed along by the circulating molten iron until the wustite forming reaction is virtually completed.

After the reduction slag separates off, the iron carrier melt, now incorporating the partly refined iron product, continues on its way for recirculation in the melt circulation loop. Continuous automatic overflow by gravity of molten iron product stream and reduction slag from the melt circulation loop takes place.

To totally avoid C0 ₂ emissions at the plant site, all the thermal energy input for the entire solid charge incorporation process can be provided by electrical conductive heating using analogous technology to that developed by Rio Tinto for the primary aluminium industry. The electrical conductive heating arrangement is comprised of two circuits with current flowing in opposite directions via two 600.000A DC contactors. All of the electric current is constrained to the molten iron carrier melt ore plus flux charging zone to provide a power input to the circulating molten iron of about 210 W maximum via cul-de-sacs off the carrier medium main flow. The molten iron circulates, within a relatively narrow open channel in this solid charge incorporation region of the ironmaking melt circulation loop. The cul-de-sacs admitting the very large electric currents are force cooled with a liquid metal coolant such as lead bismuth eutectic to permit localized freeze lining on the cul-de- sac walls. A total current input of up to of around 1200 kA DC is provided from an adjacent substation.

Pure hydrogen reduction at atmospheric pressure of a thin molten layer of oxidic melt containing liquid wustite floating on a molten iron carrier medium is almost exclusively controlled by molecular gaseous diffusion. Having been the first researcher to establish the effects of gaseous diffusion on hydrogen reduction kinetics of hematite, erroneously then promoted by researchers at US Steel as being entirely rate-controlled by chemical reaction, the inventor was naturally concerned that mixed-controlled mass transfer may dominate in the reduction zone of the presently proposed configuration. This is because of the increased level of silica in the oxidic melt resulting from the absence of any beneficiation.

In this context, detailed calculations were undertaken on magnetite ores containing 31 and 36% Si0 ₂ , which confirmed gas phase mass transfer rate control, provided

appropriate steps were taken to increase the carrier melt velocity in the reduction zone so that the oxidic melt layer thickness is reduced in thickness to ensure that liquid diffusion in the stagnant oxidic melt layer balances the reduction demand.

Because of the elimination of beneficiation, direct smelting of low-grade iron ore invariably means that the quantity of oxidic melt is far greater than that associated with earlier design assessments of the technical viability of generic melt circulation for direct smelting of high-grade iron ore such as Pilbara hematite direct shipping ore (DSO). However, recently published Magnetite Ore Resources in the 2016 Annual Report of the third largest mining company in the Pilbara, coupled with the disaster of a tailings dam failure, given huge publicity following the tragic loss of life and environmental damage in Brazil, was sufficient inspiration to undertake the current study The magnetite ore resources, just referred to, amount to an estimated total of 6,706 million tonnes with an average 31.4% iron content, with in -situ 40.27% Si0 ₂ , 2.22% Al ₂ 0 ₃ .

Direct iron ore smelting without beneficiation is the logical way of eliminating disasters associated with tailings dam failures. The most recent of which occurred in November 2015 at the BHP Billiton and Vale-owned 30 million tonne per annum operation at

Samarco, Brazil. According to a press report (The Weekend Australian March 12-13 2016), "the deadly tailings dam failure sent a tsunami of tailings down a 600km stretch of the Rio Doce to the Atlantic ocean and devasted the nearby town of Bento Rodrigues, killing 17 people and 2 listed as missing".

In the proposed process, the ore charge is mixed with an appropriate amount of flux, preferably bauxite to dismiss associated C0 ₂ emissions involved in lime or calcium aluminate production, so that the oxidic melt remains in the liquid state floating on the molten iron carrier medium throughout the entire length of the reduction zone of the ironmaking melt circulation loop. Bauxite resources in Australia include 3000 million tonnes in the Weipa region in Queensland as well as many other major locations, yielding average grades between 49 and 53% of Al ₂ 0 ₃ . In 2012 these represented 30% of global bauxite production.

For melt circulation smelting of iron ore within a melt circulation loop, electrical

conductive, heating in regions with massive un-melted shells of highly conductive solid iron in place to facilitate freeze lining should be avoided. Major energy input by

conductive heating demands a high circuit electrical resistance, which is obviously destroyed in the presence of such thick iron shells. Accordingly, conductive heating is necessarily restricted to a zone in which expensive fusion cast alumina with a protective stable layer of hercynite (FeAI ₂ 0 ₄ ), for example, can be employed rather than a freeze lining. For smelting hematite direct shipping ore (DSO) there is sufficient sensible heat in the circulating molten iron in advance of oxidic melt reduction, which is mildly exothermic, to satisfy the heat balances, taking into account a small heat loss, without reliance on electrical conductive heating in these regions, where hydrogen reduction of wustite is performed to satisfy the designed level of iron production. However, this is not the case for low-grade iron ore smelting presently under discussion, where a totally different approach is needed. Accordingly, some quantitative calculations will now be given to enhance this overall process description.

The HSC Chemistry for Windows computer program has been used to compare melt circulation direct smelting of the low-grade magnetite ore presently under discussion in comparison with zero C0 ₂ smelting hematite DSO, assuming zero CO ₂ melt circulation technology were introduced in a future low carbon economy to produce at the mine site ultra-low carbon iron slab for export worldwide.

To maintain true comparability between direct ore smelting without any beneficiation and melt circulation smelting of hematite DSO, essentially the same melt circulation

arrangement was used in both cases. The common basis is a solid charge containing one mole Fe ₃ O or alternatively 1.5 mole Fe2O ₃ to be hydrogen reduced to 2.85 mole Fe product (95 % recovery) at 1811.15 K, 1 bar total pressure by initial incorporation into 143 mole of molten iron carrier medium circulating around a closed loop employing RH gas-lift pumping technology with continuous overflow of reduction slag and molten iron product. For the production of 2Mtpa of molten iron in both cases, incorporation of non-preheated hematite DSO has a theoretical thermal energy demand of 186MW. In comparison, low- grade magnetite ore without beneficiation charged with bauxite flux, in theory, would consume in the solid charge incorporation zone about 130MW, if preheated to 1100°C using the highly efficient reduction slag thermal energy recovery scheme now being proposed.

Steps must be taken to ensure the magnetite component of the ore charge is not oxidized, so that the theoretical reduction in hydrogen consumption in comparison with hematite is realized in practice. Also, steps must be taken to compensate for the freeze lining as well the sensible heat absorbed in raising the temperature of the hydrogen reductant from say 1520°C to the reduction temperature, say 1538°C. This can be achieved preferably by electric heating elements installed above the top surface of the oxidic melt undergoing reduction. For DSO smelting this is not an issue, because of mild exothermicity. For smelting the low-grade iron ore, HSC4 indicates endothermicity, if the hydrogen is admitted at 1520°C. For example, hydrogen reduction of 1 mol magnetite with 150 mol of hydrogen involved with 31% Fe containing ore and total alumina 0.30 mol in the fluxed solid charge, preheated to 1100°C, needs a thermal input in the oxidic melt hydrogen reduction zone of about 35 MW for 2 million tonne per year of iron product, if the input hydrogen temperature is maintained at 1520°C. Alternatively the exit

temperatures of the oxidic melt as well as that of the molten iron carrier medium need to be raised to around 1560°C on leaving the electrical conductively heated ore charging incorporation zone in advance of the beginning of the hydrogen reduction zone

Illustrative Example:

Zero CO2 direct smelting a low-grade magnetite iron ore without beneficiation to produce 2 million tonnes per year of ultra-low carbon iron slab for export word wide is the basis of the following estimates:

At a mine site in the Pilbara region, Western Australia, availability of land space is considered not to be an issue, so that a thermally well-insulated melt circulation loop containing a straight reduction zone 200m in length and 15m in width is acceptable, provided thermal expansion issues and friction losses are properly taken into account. For Kelvin Helmholtz stability with a gas phase velocity of 400 m/s and a hydrogen pressure drop of less than 0.05 bar, the required reduction zone dimensions is estimated as specified.

Assuming that the ore charge plus flux is initially convectively heated by an inert gas to 460°C and then exposed to radiative heat transfer in the float glass facility as it is then preheated to 1100°C as the reduction slag cools from say 1540°C to 600°C. The estimated average radiant heat transfer intensity works out at 68 kW/m ² .and the thermal demand is about 116 MW. The exposed slag surface area floating on the molten tin pool needs to be about 1700 m ² , so a length of 190 m and a width of 9m should be adequate. For tin depth of say 5.1 cm, the minimum total mass of tin required is 553 tonne, so the capital cost is expected to be in the region of 15 million US $.

A well known electrical contractor in Europe has generously supplied the estimated capital cost of the electrical conductive heating to satisfy a demand of 210 MW in the present context as follows:

12 pieces regulating/rectifier transformers, assumption prim, voltage 132kV

12 pieces diode rectifiers, output lOOkAdc, 0-175Vdc

12 pieces unit control

1 piece master control

1 piece micro SCADA

To provide 1200 kA DC; 12 units, each rated at 100 kA, DC is required. The assumption is that the units should be rated for 175Vdc and that the total power consumption is 210MW.

The estimated price (non-binding, but indicative for commercial evaluation of the proposed process) is 72 MEUR.

The estimated price (non-binding, but indicative of the requirement) not including any civil works, installation work, bus bars to electrically connect the units to the process, etc.

An embodiment of the present invention will now be described, by way of example only, with references to the accompanying diagrams, in which:

Figure 1.is schematic-sectional plan view of the principal ironmaking melt circulation loop with a single large RH gas-lift pumping facility.

Figure 2 shows the arrangement for charging preheated ore plus flux along the whole length of the electrical conductive heating region to effect incorporation of solid charge with minimal temperature rise of the molten iron carrier medium.

Figure 3 is a cross-sectional elevation of the vitally important oxidic melt hydrogen reduction zone. Figure 4.is a schematic cross-sectional elevation of the enhanced safety provision of low pressure steam enshrouding the close to atmospheric pressure hydrogen reduction oxidic melt reactor detailed in Fig. 3.

Figure 5 illustrates the basic mechanism for massive radiative heat transfer between very high temperature reduction slag and ore plus bauxite flux undergoing preheating, on which the viability of the new technology for elimination of beneficiation is absolutely essential.

Figure 6 is an acknowledgment that prior art provides the most desirable solution to meeting the objective, stipulated as essential in Figure 5 by an adaptation of the commercially well-proven float glass process.

The proposed overall ironmaking process is shown schematically in Fig. 1 , approximately to scale in sectional plan view. For the base case of 2 Mtpa Fe, the overall length of the melt circulation loop 1 is estimated to be approximately 180m. The left-hand end of the loop 2 is anchored, whilst the rest of the loop expands freely to the right. This means that substantial lateral movement mounts up progressively along the length. For example, the 600,000 amp DC current connectors 3 and 4, which input the currents to what are effectively "cooled cul-de-sacs" 5, 6, 7 and 8, incorporated into the main current flow paths 9 and 10, must have built-in flexibility employing, for example, short vertical 20cm diameter vertical steel posts (not shown), partially hollowed out to permit access of a steel pipe introducing a stream of recirculated liquid metal coolant, probably molten lead- bismuth, at say 12 adjacent sites (6 in each of the two cul-de-sac current input side-by- side locations). The liquid metal coolant then proceeds in the hook-shaped steel post to a horizontal external launder adjacent to a copper bus bar also immersed in the circulating liquid metal coolant to minimize contact resistance, whilst overcoming adverse effects of differential thermal expansion.

Also the RH snorkels require cooled rubber or plastic coated bellows (not shown) to permit free up and down movement to prevent air ingress into the protective atmosphere in contact with the molten iron 11. If the steel posts referred in the previous paragraph are relatively short, say 1m in length, the overall power loss over the 12 electrical connectors is estimated to be just over 5MW. This means that the transmission of electrical energy from the substation to the circulating molten iron can be anticipated to be about 95% efficient.

The two 600,000A DC current connectors 3 and 4 positioned in the cooled molten iron cul-de-sac arrangements 5 and 6 transmit the major portion of thermal energy demands involved directly in the overall chemistry associated with reducing magnetite to metallic iron at a rate of 2 Mtpa at 95% Fe recovery.

A decision has to be taken at the design stage on whether or not to use expensive fusion cast alumina blocks and segments. The full implications will need detailed financial assessment and justification. It is conceivable that ordinary high alumina bricks could possibly be used throughout the melt circulation loop, provided the interfacial regions in the hydrogen reduction zone and the prior solid charge incorporation zone are protected with a thin strip or longitudinal thin freeze lining. As a first estimate, the capital cost in the base case of 2 M tpa product for fused cast alumina blocks or segments backed up by high alumina ceramic board delivered to the site by a manufacturer is about 53M US $.

As shown in Fig. 1 , concerning phase disengagement associated with the low intensity of this whole reduction zone and as the adjacent snorkel inlet is from the bulk of the flowing molten metal well underneath the slag/metal interface, additional phase separation is not required for removal of reduction slag at 14. Molten iron product 15 overflows from the principal melt circulation loop into a steel refining melt circulation loop in advance of continuous casting of steel product, or alternatively cast straight away into product ingots, if undertaken at the mine site for export. The hydrogen reductant is not shown in Figs.1 or 2, because both the hydrogen input and the reduction off-gas ports are well above the section illustrated and are usually placed on the top of the oxidic melt reduction zone with the gas flow moving counter current to the extensive stable floating oxidic melt layer, which spatially completely dominates the plan view in Fig. 1. This floating layer moves forward at virtually the same velocity as that of the molten iron carrier medium. Figure 2 is to be observed in conjunction with the details already given in Fig. 1. The major difference is that in Fig. 1 the solid charge location 13 is conducted at a single location immediately after the molten iron carrier medium leaves the electrical

conductively heated molten iron carrier medium at an excessively high temperature with the consequence of major heat loss to the freeze lining and thus increased energy demand on the conductively heated region. In Fig. 2 multiple locations stem from 13. The other figures 14 and 15 result from the ore charge, with subsequent hydrogen reduction of the thin oxidic melt layer to the final product 15 molten iron overflow and the reduction slag 14, also overflowing continuously.

In Fig. 3, features are illustrated to accommodate differential linear thermal expansion between the roof arrangement etc and the refractory-lined or freeze lined hearth and side walls (16a, 16b & 16). In the former case, the all important fused cast alumina blocks or segments 16a are in contact with the circulating molten iron 15a. High alumina refractory board 16b is in close contact with the fusion cast alumina 16a and provides a degree of thermal insulation in advance of the low alloy steel encasement, which by design is no hotter than about 600°C and controlled at such using forced cooling either by steam raising boiler tubes 22 or alternatively by forced convective heat transfer to an inert gas re-circulated externally for heat recovery to the solid charge of iron ore plus flux for drying purposes or initial preheating, for example. This is principally because of the sensible heat transported by the molten iron carrier medium to the mildly exothermic reduction zone by the circulating molten iron carrier medium 15a. The hydrogen gas space 17 separates the oxidic melt layer 18 from the flat refractory roof 19 supported by steel joist girders 20. Differential thermal expansion between the roof arrangement containing steel joist girders and the open-channel melt circulation loop is facilitated by provision of support pontoons 23 floating on a dense low-temperature fusible alloy such as lead- bismuth eutectic or possibly molten lead. Segmented removable thermally well-insulated lids cover the whole gas-tight enclosure. In Fig.3, a fibrous ceramic board 16 seals in the roof structure once it reaches its stabilized position. It is then no longer dependent on the pontoon support arrangement 23. When discussing the positive advantage of generic melt circulation technology over other smelting reduction technologies such as Hi-smelt, recognition of the virtue of low reaction intensity was highlighted in the ore incorporation zone. The case continues now in the oxidic melt reduction zone, where the stability of the hercynite protective lining adjacent to the oxidic melt floating layer is crucially important on both side walls illustrated in Fig. 3. Again, a thin strip or fin of freeze lining is mandatory at location 24 shown on both sides of the open channel extending from the melt through to the forced cooled region. However, the temperature driving force for heat loss to the freeze lining is considerably less than in the ore incorporation zone but on the other hand the estimated length of the channel is 141m compared with 35m for ore incorporation.

Still referring to Fig. 3 hydrogen is forced to flow at high velocity (around 400 m/s) below the critical velocity, defined by the Kelvin-Helmholtz relationship for interfacial instability, through the confined space 17 between the top surface of the oxidic melt and the roof or ceiling of the reduction. The reduction off-gas (not shown) then proceeds directly through a refractory-lined duct (not shown) to a pebble regenerative heat exchanger, known as a "pebble heater" (Stevanovic, D. and Brotzmann, K. 2004. Pebble-Heater Technology in Metallurgy, Metalurgija, 10 (1) 19-36). After leaving the pebble heater, the gas is eventually cooled to condense out the water vapour and effect gas cleanup by filtration.

A really significant aspect of the current proposed reduction process involving major presence of gangue, principally silica, stemming from deletion of beneficiation, as opposed to reduction of as- mined high grade iron ore, is the increased amount of hydrogen needed to be recirculated. Fortunately thermodynamics predicts the actual amount consumed by low-grade magnetite is virtually the same as for reduction of high grade magnetite concentrate. Unlike a CO reductant, the ratio H ₂ /H ₂ 0 mixtures on cooling the reductant gas remains constant. Hence recycling is facilitated without increasing ultimate reductant consumption.

For a typical Pilbara (Western Australia) low-grade magnetite iron ore (31% Fe) some 150 mol H2 needs to be added to the reduction region in a melt circulation loop to induce at steady state a molten iron continuous overflow of 2.74mol iron product per mol Fe304,corresponding to 93.1 % recovery of iron units charged. For a direct shipping hematite ore (DSO), the corresponding figure for 95% recovery of iron units charged is a mere 16.3 mol H ₂ for a charge of 3 iron units, corresponding to 1.5 mol Fe ₂ 0 ₃ . However, as to be expected from thermodynamics, reduction of magnetite consumes less hydrogen reductant than reduction of hematite. For the production of a single mol of metallic iron product some 1.365 mol H ₂ is consumed for direct smelting of the low-grade Pilbara magnetite ore, (31% Fe) whereas 1.528 mol H ₂ consumption is required to smelt the Pilbara DSO.

Figure 4 highlights the enhanced safety aspect of operating close to atmospheric pressure. The reduction zone gas-tight enclosure 27 is encased in a horizontal cylindrical steel vessels 28 containing low pressure steam 29, totally surrounding hydrogen- containing plant items, including the thermally well-insulated lift-off lid arrangements 30.

Referring now to Fig. 5 this shows the gas-tight refractory lined thermal radiation enclosure 31 in cross-sectional elevation. Immediately adjacent side-by-side is the top surface of the slag 32 is adjacent to the horizontally conveyed solid charge strand or layer

33 undergoing preheating. Throughout the whole radiative heat exchange process the slag layer floats on a static bath of non-wetting molten tin in an open channel. The saturated vapour pressure of molten tin at1500°C is approximately a mere 5x10 ^"4 bar and at 600°C is 5.5x10 ^"13 bar, so evaporation is not a problem. Serth et al, the original proposers of recovery of waste heat from industrial slags via a modified float glass process, recommend a protective gas atmosphere 92% 4% H ₂ 4%CO to prevent the tin

34 from oxidising.

In addition to the anticipated reduction in operating and capital costs, the proposed continuous ironmaking could pave the way towards future autonomous refined iron production at a mine site from iron ore containing high phosphorus content. This would facilitate viable exploitation of at least 8 billion tonnes of additional Australian iron ore resources. A major challenge facing the larger mining companies, particularly in Western Australia, is how to reduce labour costs introduced by the fly-in, fly-out lifestyle. Ironmaking process described in this current patent could readily be fully automated and in the mine of the future be controlled along with the iron ore mining operation itself at a central location, such as Perth, to support autonomous mining and primary steel production in the Pilbara region. In this case, the by-product carbon pellets in the proposed new technology based on thermal decomposition of natural gas to supply the hydrogen reductant could be transported to the coastal region using existing rail infrastructure along with refined iron slab product.

The by-product carbon pellets from natural gas-based hydrogen preparation at the mine site could constitute one of the charge materials in a proposed carbothermic aluminium process (Warner, N. A. 2008, Conceptual design for lower-energy primary aluminum, Metall. and Materials, Trans. B, 39B, April, 246-267). In this case, C0 ₂ disposal offshore is one option, probably associated with the existing Gorgon Carbon dioxide Capture and Storage (CCS) project. To maintain zero gas emission iron production back at the mine site, lime and calcium aluminate fluxes could also be prepared at the CCS hub and then transported along with the desired bauxite flux to the mine site using the existing rail infrastructure.

Maintenance of a low gas phase pressure drop is the most important factor in securing plant operation close to atmospheric pressure. The over-riding requirement of enhanced safety is facilitated by these means. In addition, the mass of hydrogen at high

temperature held-up within the process chain is reduced to an absolute minimum.

Also significant in ensuring safe operation is the provision of a low- pressure steam sealing arrangement in containment vessels or jackets surrounding the key plant items. This would prevent hydrogen escaping into the surrounding atmosphere. In the event of malfunction or incipient plant failure, steam will flow into the hydrogen circuit, diluting it to such an extent that chemical reactions will cease and the plant automatically shut down. A range of technical papers have already been published outlining proposed generic melt circulation technology for truly continuous iron and steelmaking. In the present invention, the all Important emphasis is on energy recovery from slag at high temperature to secure the benefits of elimination of beneficiation in environmentally friendly direct smelting of low-grade iron ore.

The final thin solid sheet of slag is pulled through the heat recovery operation by pinch rolls after separation from the molten tin is accomplished. Any tin droplets are removed by an "air knife" employing nitrogen or other inert gas rather than air. It is worth noting that molten tin cannot be used In continuous casting of the ferrous metal product to eliminate water cooling. The solubility of tin in molten iron is far too great for this to ever become a realistic proposition.

Radiative counter-current heat transfer from the newly formed slag layer to the adjacent conveyed solid charge takes over immediately, whilst floating on a static bath of molten tin. Clearly, this is a novel adaptation of the globally successful float glass process.

Recovery of heat from molten slags in the steel, copper and elemental phosphorus industries using modified float glass technology was proposed by Serth et al. (Energy Communications, 7(2), 167-188 (1981), R W Serth, T E Ctvrtnicek, R J McCormick and D L Zanders) in research to determine the technical and economic feasibility of energy recovery of an energy recovery facility generating steam exported from the system. For electricity generation the return on investment was found to be negative. The float chamber containing the molten tin was by far the major capital cost identified.

The research company BMI Research in November 2016 forecasted the tin price in 2018 to be $20,500/t and commented on the stronger global demand for the metal compared to production growth. Clearly, substantial future tin consumption in the iron and steel industry may cause a dramatic increase in the price of tin, so an alternative option for slag energy recovery in the present context should be given at least some consideration.

For example, a circulating fluidised bed (CFB) and turbulent fluidised bed heat exchange unit was constructed to demonstrate the feasibility of an advanced slag heat recovery scheme at the University of Birmingham. This demonstrated the feasibility of producing slag droplets using self-impinging jets of molten slag (Redcar iron blast furnace) (CP Broadbent, and N.A. Warner: 1984, Trans.lnstn.Min.Metall, Sect. C: Mineral Process. Extr. Metall.), vol. 93, C130- 33. and that these can be cooled successfully in a CFB (N.A. Warner, T.J. Peirce, J.W. Armitage, J.S.M. Botterill and Y. Sergeev: 1994 Tenth

International Heat Transfer Conference, Paper No. 64). Short term tests showed a response consistent with a droplet-to-CFB heat transfer coefficient in excess of 400 W/m ² K when interpreted on the basis of a simplified unsteady state model, which assumed that the CFB and turbulently fluidized section can be represented by separate stages in series.

Consideration of potential steady-state heat balances illustrated the control exercised by the ore or overburden fines flow rate through the CFB in regulating the temperature of the turbulent bed. The flow characteristics of large particles of slag counter to a CFB need to be determined experimentally so that the necessary height of the CFB to achieve the required fall time through the CFB heat exchanger can be established. To be able to reach steady state operations and verify the magnitude of droplets-to-CFB heat transfer coefficient, it is necessary to set up an experimental rig in an industrial plant, where adequate supplies of molten slag are available.