Oxygen-Enriched Bottom-Blown Furnace

What is an oxygen-enriched bottom-blowing furnace and how does it differ from a side-blown or top-blown furnace?

An oxygen-enriched bottom-blowing furnace is a bath-smelting reactor in which oxygen-enriched gas is injected through submerged lances set in the furnace floor, so the oxidant enters the matte layer at the bottom of the bath rather than being blown across the slag surface. The difference between the three architectures is the gas injection position and nothing else, and everything that follows from it. In a top-blown or side-blown vessel the oxygen has to reach the reaction site through the slag and the bath surface, so the vessel is organised around keeping that surface violently disturbed, the oxygen utilisation is limited by how much of it reaches the metal, and a slag that is thick or viscous or rich in magnetite obstructs the reaction. In a bottom-blowing vessel the gas is delivered where the sulfide is, so most of the oxygen reacts in the matte, very little reaches the slag, less magnetite forms, the slag stays fluid and the bath separates. The published operating figures reflect it: oxygen utilisation approaches 100 per cent with bottom injection against roughly 30 per cent for a top-blown rotary converter, and the reaction rate per unit of furnace volume runs at ten to twenty times a reverberatory furnace.

Why is the oxygen blown into the matte layer instead of into the slag?

Because the matte layer is where the sulfide is, and the sulfide is what the oxygen is there to react with, and because putting the oxygen into the slag instead produces magnetite rather than metal. When oxygen contacts an iron-bearing slag it oxidises iron to magnetite, Fe3O4, which raises the slag viscosity and can build a solid layer between the matte and the slag that stops the two phases separating. Delivering the oxygen below the slag avoids most of that formation, and the consequences chain outward: the slag stays less viscous, no isolating layer forms between matte and slag, matte droplets settle out of the slag instead of being held in it, a high-iron slag type becomes usable, and a high-iron slag needs less flux and produces less slag per tonne of feed. Two further benefits follow from the same cause. Because the slag layer does not have to be penetrated for the reaction to proceed, a thicker slag can be carried, and a thicker slag holds a smaller proportion of the interface metal relative to its volume, so less metal leaves with the tap. And because the injection point is below the bath surface, the bath does not have to be kept in a violently splashing condition to keep the oxygen in contact with the metal, so the process is not dependent on holding a broken surface.

What matte grade can the furnace make, and what does a higher grade cost?

The matte grade is engineered to the step downstream and is normally made in the range of about 45 to 70 per cent copper, and the cost of a higher grade is paid in the slag. Copper loss to the slag rises continuously as the matte grade rises, and the rise becomes steep once the matte passes roughly 68 per cent copper, because at higher grades the slag carries more copper and the matte droplets in it are both more numerous and more valuable. A high-grade matte is worth more per tonne, it carries less iron through to the converting step and it reduces the load on the converter, so the choice is not obvious and it is not a technical question alone. We present the matte grade, the copper content of the slag and the duty that the slag-cleaning circuit has to carry as three numbers from one calculation, so that the decision is made with its penalty visible. A furnace quoted at a single high matte grade without a slag loss figure attached to it is half a specification.

Why does this furnace produce less magnetite in the slag than a top-blown or side-blown furnace?

Because the oxygen is delivered below the slag layer and reacts with the sulfide in the matte before it can meet the iron in the slag. Magnetite, Fe3O4, is formed when oxygen contacts an iron-bearing slag, and in a reactor where the oxidant has to pass through the slag to reach the metal that contact is unavoidable and continuous. In a bottom-blowing furnace the lances discharge into the matte layer, so the oxygen meets sulfide almost immediately and is consumed there, and very little of it crosses into the slag. The slag therefore operates at a lower oxidising potential than the slag in a top-blown or side-blown vessel, and less magnetite is generated in it. That is worth more than it sounds, because magnetite is the mechanism behind several of the classic bath-smelting problems: it thickens the slag, it slows the settling of matte droplets, it can build a solid layer at the matte-slag interface that stops separation, and a slag loaded with it has to be made thinner with more silica, which means more flux and more slag and more metal lost per tonne of feed.

What is the mushroom head on the lance and why does it matter?

The mushroom head is a solid accretion of frozen bath that forms over the outlet of a submerged lance under the right combination of shroud gas flow, gas velocity and immersion depth, and it is the reason a lance can survive being immersed in liquid material at over 1,000 degrees. It works two ways at once. It shields the lance tip and the lance body from the reaction that is happening immediately around it, which is why the lance is a wear part with a life measured in months rather than in hours. And it holds the region of highest reaction rate out in the bath, away from the refractory brickwork, so that the brick is not washed by the most aggressive part of the process. That second function is why the wear in a properly run bottom-blowing furnace concentrates at the slag line, where the attack is chemical and where the answer is the refractory grade, rather than around the lance, where the answer would be a shorter campaign. The accretion is not luck: it is produced by the shroud flow, the velocity and the immersion, which is why commissioning the lance set is a procedure with a target condition rather than a matter of inserting the lances and starting up.

How is the lance set designed, and does the number of lances matter?

It matters, and so do the angles, the spacing, the stagger and the immersion, which is why the lance set is issued as an engineering drawing rather than as a nozzle count. A single lance or a small number of closely spaced lances delivers the oxygen into a few local channels and stirs the bath near those channels rather than agitating it as a whole, which leaves part of the bath under-reacted and produces local hot spots. Published bottom-blowing copper practice shows the level of detail the design reaches: eleven oxygen guns set in the furnace bottom in two rows, five in the lower row at 7 degrees and six in the upper row at 22 degrees, with 15 degrees between the rows, and with the two rows staggered so that the gas streams from the upper row fall between the streams from the lower row rather than on top of them. Number, angle, spacing and stagger are calculated together against the bath diameter and the oxygen flow, because they decide the distribution of the oxygen potential across the bath floor, and the immersion is held by the advance mechanism so that the injection depth stays where the design put it.

Can the lance be replaced without draining the furnace?

Yes. The lance set is arranged so that a lance can be brought above the bath surface by tilting or rolling the furnace, and at that position the lance can be withdrawn and replaced without any of the content draining out through the lance opening in the lining. That is what makes the lance a planned consumable rather than an unplanned outage: the change is scheduled into the campaign, a spare set is held on site, and the inspection of the worn set is where the plant learns whether the shroud flow and the immersion on each position have stayed where commissioning set them. Uneven burn-back across a set, or a set that has worn at different rates from position to position, is the signature of a shroud flow or an immersion that has drifted, and it is correctable before the next campaign rather than at the next failure.

What feeds can the furnace accept, and does the feed need to be dried or granulated?

It accepts a wide and deliberately wide range, and it does not require drying or granulation where the moisture is inside the window. Copper sulfide concentrates at natural moisture, copper oxide and mixed oxide-sulfide material blended with sulfidic feed for its reaction heat, galena and lead-zinc bulk concentrates, lead-acid battery paste raw or desulfurised, zinc plant leaching residue, copper scrap and cable, printed circuit board concentrates and electronic residue, pre-treated spent catalysts, anode slime, flue dust, filter cake and mixed metallurgical residues, off-specification matte, and precious-metal-bearing bullion and doré feed. Feed with a moisture content up to about 8 per cent can be charged directly, and the material does not have to be granulated or deeply dried, which removes the dryer, the granulation circuit and their fuel and maintenance from the flowsheet. Lead-bearing material has been run at grades as low as about 30 per cent. Where the feed is variable or mixed, it is blended to a designed copper, lead, sulfur, iron and precious-metal content and to a designed moisture, so that the heat balance, the matte grade and the slag composition all land inside the window the furnace was built for.

Does the furnace need fuel, or does it run on its own reaction heat?

It runs on its own reaction heat for the bulk of a campaign, and it is designed with a burner for the parts it cannot cover. Oxidising a metal sulfide releases a large amount of heat, and with oxygen enrichment that heat is enough to hold the bath at temperature and to melt the incoming charge, so the furnace operates in a fully autothermal mode. The supplementary burner is sized for heating the charge from cold at the start and for holding the bath while the furnace is being cast or taken down; it is not sized to carry the process heat. Whether the furnace runs fully autothermal or semi-autothermal depends on the sulfur and the iron in the feed and on the moisture and carbonate that consume heat, so the mode has to be switched when the feed changes. That is why the balance is calculated and issued with the burden recipe: autothermal capability is a balance to be held and not a property of the vessel, and a furnace sold as autothermal and then run on a feed that cannot hold the bath will be blamed for a control failure it did not cause.

Why is the furnace shell insulated instead of water-jacketed?

Because a water jacket maintains a permanent path from liquid material above 1,000 degrees to water, and the failure mode of that path is a steam explosion rather than a leak. An insulated shell lets the refractory lining carry the containment duty, with a frozen layer of bath held against the brick by the heat balance rather than by a cooling circuit, so the molten material is kept away from the steel shell by design instead of by the integrity of a cooling water system. There is a thermal benefit as well, since heat that is not taken away by cooling water does not have to be supplied by the reaction or the burner, so an insulated shell holds the bath temperature on less fuel. But the reason the choice is made is containment, and it is worth stating plainly because the water-jacketed alternative is common in other furnace families and is sometimes offered on bath-smelting furnaces without the failure mode being discussed.

What happens to the slag, and how do you keep copper or lead out of it?

The slag is tapped from a slag taphole at the settling end of the vessel, where the matte droplets carried over from the reaction region have had the residence time to settle back into the matte rather than leaving with the slag, and the tap is controlled against an assay rather than a level. The slag metal content is what the recovery of the plant turns on, so the slag is sampled at every tap and checked against a release figure or a slag-cleaning figure. It is kept low in three ways. Mechanically, by delivering the oxygen below the slag so that less magnetite forms, the slag stays fluid and the droplets settle, and by carrying a deeper slag so that the proportion of interface metal relative to the slag volume falls. Chemically, by holding the slag composition on the target that the feed and the flux were calculated for, since a slag off its target composition holds more metal in solution. And operationally, by treating the matte grade and the slag metal content as one decision: the higher the matte grade, the more metal the slag will carry, and where the furnace is run at a high grade the slag is tapped as a stream for slag cleaning rather than as a finished discard, with the recovered values returned to the burden.

Is one furnace enough, or does the process need two?

It depends on the route. For a copper feed one furnace is normally the smelting vessel, with slag cleaning handled in a separate settling or flotation step. For lead-bearing and zinc-bearing feed the established layout uses two: the feed and its flux go to the oxidation furnace, which produces lead bullion and a high-lead slag, and that high-lead slag goes to a second bottom-blowing furnace where a reductant is injected into the bath and the oxidised lead and zinc are reduced back to metal. The reduction furnace produces lead bullion, which rejoins the product stream, and a zinc slag, which goes on to a fuming furnace where the zinc leaves as an oxide powder recovered from the fume. Two vessels let each be designed around one duty, one slag type, one temperature and one oxidising or reducing regime, instead of one vessel alternating between regimes that want different conditions and being mediocre at both. The off-gas from the oxidation stage is rich enough in sulfur dioxide for an acid plant, while the gas from the reduction and fuming stages is desulfurised by absorption.

What happens to the off-gas, and can the sulfur be sold?

The off-gas leaves with sulfur dioxide, dust, heat and whatever volatile metals the bath has thrown off, and all four are handled before the gas goes anywhere. It passes to a waste-heat boiler first, where it is cooled to a temperature the downstream equipment can accept while the heat is recovered as process or power steam, and where the gas residence time in the high-temperature zone is designed to be long enough for the carbon monoxide and the organic fraction to combust fully before cooling, which is the control on dioxin formation where the feed carries electronic or plastic material. The dust is then captured. On a sulfidic feed the cleaned gas goes to an acid plant, and it can be sold as acid because the oxidation stage runs continuously, so the sulfur dioxide content of the gas sits in a stable band rather than swinging with a batch cycle. Stability is what an acid plant needs: a gas flow and strength that move around cannot be run into a saleable product. Lead bottom-blowing practice reports the sulfur dioxide content of the oxidation off-gas in the 12 to 18 per cent region. Where no acid plant is fitted, or where the gas is too weak, as it is from the reduction and fuming stages, it is desulfurised by absorption before release.

What happens to the dust collected from the gas?

It is captured, weighed, assayed and returned to the burden or routed to its own recovery step, because on a lead, zinc, copper or precious-metal line the dust carries payable metal and is one of the richer streams in the plant rather than a waste. The dust load is a design output of the process calculation, the collector is sized on the off-gas volume and that load, and the emission figure at the stack is specified against the limit that applies at your site with the collection efficiency selected to meet it. The dust is also a term in the campaign metal balance, so what leaves in the gas and what comes back from the collector are both recorded; a plant that treats its baghouse catch as a disposal item is discarding metal that it already paid energy to volatilise.

How long do the lance and the refractory last?

Under the designed injection pattern the lance is a wear part with a life measured in months, and it is changed at a planned point in the campaign rather than on failure, with a spare set held on site. The refractory lining is a campaign item measured in years, and its length is not a calendar figure: it is set by the wear at the slag line, where the slag and, in lead-bearing service, lead oxide attack the brick, and where the reference grade is a fused-grain, direct-bonded magnesite-chrome brick. Because bottom injection holds the highest-rate reaction zone away from the lining, wear is concentrated at the slag line rather than distributed over the vessel, which is what makes a long campaign possible and what makes the slag line a detail that can be measured at every shutdown and replaced as an element. The measurement we ask the plant to keep is a wall-thickness and lining-thickness survey against the original drawing and against the previous survey, so the wear trend is a curve the plant manages rather than a number discovered at failure.

What utilities and site services does the furnace need?

Electric power for the drives, the control system, the dust collector and the gas train; the oxygen supply, either as an on-site vacuum swing adsorption or pressure swing adsorption plant or as liquid oxygen storage with a vaporiser; instrument air and a shroud or purge gas supply, which is normally nitrogen; cooling water and boiler feedwater with its treatment, since the waste-heat boiler produces steam and needs a feedwater quality behind it; granulation or casting water where the tapped matte or slag is granulated; and fuel for the supplementary burner, which carries charge heating and bath holding rather than the process heat. On the gas side, the equipment is a function of the route: a waste-heat boiler, a dust collector, gas cooling, and then either the acid plant or the desulfurisation train. On the material side, weigh feeders and bins for concentrate, flux and reductant, and ladles, launder systems or granulation and handling for the matte, the bullion and the slag. All of those quantities are calculated from the mass and heat balance for your feed and issued as a service schedule rather than quoted as a standard set.

What is tested at acceptance, and how is the recovery verified?

Acceptance is run as a performance test on your feed and it is measured as a metal balance, not as a duration. The line is operated at the designed charging rate to demonstrate the designed matte grade, the agreed slag metal content and the throughput, and during the test every stream is weighed and assayed: the feed and its flux and reductant, the matte or bullion, the slag, the dust and the gas. The balance is then closed on feed in against product plus slag plus dust, with the residual metal drained at shutdown and the spent refractory carrying attached metal also accounted for, and the difference is reported as the loss for the test rather than absorbed. The lance set is checked at its designed angles and immersion with the per-lance flows balanced and recorded, the advance mechanism is verified against its reference, and the control and interlock schedule is proven, including the trips for loss of shroud flow, loss of oxygen pressure, high bath temperature, high gas temperature at the boiler inlet and loss of feed. The performance test report, the balancing record and the closed metal balance are handed over with the plant, so that the recovery figure is one you can audit and reproduce.