Recovery of iron and zinc from blast furnace and basic oxygen furnace dusts: A thermodynamic evaluation

Minerals Engineering, Vol. 7, No. 8, pp. 985-1001, 1994 Copyright ~, 1994 Elsevier Science Lid Printed in Great Britain. All rights reserved 0892-6875/94 $7.00 + 0.00

Pergamon 0892-6875(94)00043-3

RECOVERY OF IRON AND ZINC FROM BLAST FURNACE AND BASIC OXYGEN FURNACE DUSTS: A THERMODYNAMIC EVALUATION

S.M. HAY§ and W.J. RANKIN't § Mount Isa Mines Limited, Private Mail Bag, Mount Isa, Qld 4825, Australia G.K. Williams Cooperative Research Centre for Extractive Metallurgy, University of Melbourne, Parkville, Vie 3052, Australia (Received 1 July 1993; accepted 24 September 1993)

ABSTRACT Large amounts of dusts are produced from the blast furnace and basic oxygen furnace. These contain high levels of iron but their zinc contents of up to 1% and 1-6%, respectively, may prevent their being recycled to the blast furnace or basic oxygen furnace and, at present, they are often dumped. A thermodynamic modelling study was undertaken to identify the types of processes that could be used to treat these dusts pyrometallurgically to remove the zinc (and other volatile elements) and return the iron as hot metal, metallised clinker or iron oxide clinker to the smelting jqowsheet. Particular attention was given to the possible use of bath smelting processes under development as well as kiln processes as presently used to treat electric arc furnace dusts. The smelting options examined were single-stage smelting, smelting with pre-reduction, smelting with post-combustion and smelting with pre-reduction and post-combustion. Solid-state reduction to produce FeO-rich clinker and metallised clinker was also considered. It was concluded that a range of operating regimes is feasible and that the choice of process will depend on local circumstances. Keywords Blast furnace, basic oxygen furnace, steelworks, dusts, hot metal, clinker, zinc, bathsmelting, pre-reduction, post-combustion, smelting, recycling.

INTRODUCTION Over ten million tonne of iron blast furnace (BF) and basic oxygen furnace (BOF) dust is generated annually world-wide. Because of the high tramp element content and the fine nature of the dusts much of this material is stock-piled or dumped in landfill sites. This represents a loss of iron and is also a potential environmental problem due to leaching of heavy metals. BF dusts usually contain less than 1% zinc and 30 - 40% iron and BOF dusts contain typically 1-6% zinc depending on the type and amount of scrap used and 55 - 65 % iron. Recycling of the dusts to the blast furnace is the most common method of treating dusts but this is not possible for the higher zinc-containing dusts because an outlet for zinc from the system is necessary. A few processes have been developed to treat these relatively low zinc dusts though many more processes have been developed to treat electric arc furnace (EAF) dusts, which contain much higher zinc contents, because recycling to the blast furnace is not feasible and landfill disposal of EAF dusts is banned in many countries. Most of these processes do not recover iron but rely on the sale of a zinc-rich fume for economic viability. ME 7/8-.-42

985

986

S.M. HAYand W. J. RANKIN

Pyrometallurgical and hydrometallurgical approaches are possible for treating BF and BOF dusts, though the hydrometallurgical option is viable only when the zinc is present almost entirely as oxide and/or metal [1] since the presence of zinc ferrite and silicates makes leaching of zinc difficult. The present investigation was restricted entirely to pyrometallurgical processing. A thermodynamic evaluation was undertaken to identify possible processing routes for BF and BOF dusts which would recover the iron and remove the zinc. Particular emphasis was given to treating these dusts by bath-smelting processes in which heat is generated in a metal or slag in the bath by submerged combustion of a fuel and the dusts are added to the bath either directly or after some pre-heating and pre-reduction. Several processes for treating iron-bearing materials are nearing comrnercialisation, including the Australian Hlsmelt process and the Japanese DIOS process [2,3]. The Australian SIROSMELT process [4] developed for smelting non-ferrous metals, may also be suitable for smelting BF and BOF dusts if it can successfully operate at the higher temperatures required. Piret and Miiller [1] predict that direct reduced iron produced from low-zinc steelworks dusts will command a lower price than iron produced from virgin ore due to the presence of residual impurities, particularly alkali metals. They state that with economic profitability unlikely, the main criteria for the acceptability of a treatment process by industry are: •

generation of an iron-rich product suitable for use within the steel industry;

•

acceptance of the zinc-bearing sub-product by a non-ferrous smelter; and

•

generation of non-hazardous wastes.

The capability of the process to remove deleterious elements besides zinc from the iron product would increase the chances of economic profitability. The thermodynamics of reduction of pure iron oxides to form solid iron and molten hot metal is described in numerous standard texts and will not be considered here. The approach adopted in this work was to apply the Gibbs free energy minimisation technique to calculate the equilibrium composition of multicomponent, multi-phase, reactive systems relevant to dust smelting. Modelling was performed using the CSIRO Thermochemistry System (Version 5.1) with the Chemix and Model programs [5]. Thermodynamic data for all the species were retrieved from the databases of the Thermochemistry System. Blast furnace dust consists of small unmelted particles of the charge components to the blast furnace carried away in the furnace off gas. BOF dust forms mainly from melted material ejected by the high velocity oxygen jet. Iron is present in BF dusts mainly as hematite and as hematite, magnetite, wustite and metallic iron in BOF dusts. The compositions of the dusts examined in this study are given in Table 1.

SMELTING TO PRODUCE HOT M E T A L In submerged-combustion smelting, or bath-smelting, processes, fuel and air or oxygen are introduced into the bath (slag or metal) where the fuel is combusted. A major advantage of bath smelting is the high heat and mass transfer rates that are obtainable as a result of the high level of turbulence of the bath. Additionally, the bath provides a reducing medium for iron oxide usually by addition of carbon in the form of coal. Bath-smelting can be combined with post-combustion of the off-gases above the bath to recover some of the chemical energy of the gases as heat in the bath. Alternatively, or additionally, the off-gases can be used to pre-heat and pre-reduce the raw materials before they are charged to the bath. Conceptually, bath-smelting can be performed at two extremes of operation, one in which the combustion and reduction zones in the bath are fully separated, and the other in which the zones are fully mixed. The benefits in keeping the zones separated can be substantial in some situations and, therefore, the effect of degree of inter-zonal mixing was examined in some detail. It is not the purpose of this study to consider

Recoveryof iron and zinc from furnacedusts

987

the design of reactors and how the degree of inter-zonal mixing can be controlled but to identify the important process parameters that need to be considered in the design of reactors. TABLE 1 Compositions of Blast Furnace and BOF Dusts Used in the Ih'e~ent Investigation.

Species

Blast furnace dust (%)

BOF dust (%)

Total Fe

40.8

55.4

Metallic Fe

0.1

4.3

Fe H

3.1

28.1

Fem

37.6

23.0

SiO 2

6.1

2.3

AI203

2.2

0.17

TiO z

0.35

0.19

P2Os

0.10

0.21

MnO

0.58

1.8

CaO

7.4

10.6

MgO

1.9

3.7

Na20

<0.05

0.29

K20

0.08

0.22

ZnO

0.06

1.7

S

0.14

0.05

C

20.6

N/A

In the thermodynamic models, dust species and flux entered the system at 25°C as single, solid phases at unit activity, as did the carbon reductant. Apart from Fe, C and S all dust species were assumed to be simple oxides. Species chosen to define the gas phase were CH 4, 02, CO, CO 2, H2, H20, H2S, Zn, SiO, Mn, Na and K. The slag was assumed to consist of oxide species from the dust plus C.aS and Ca3(PO4) 2. The metal consisted of carbon and all elemental species originating from the dust apart from oxygen. The Raoultian activity coefficients of iron oxides in the slag and of carbon and iron in the metal were assigned the value of one. The activity coefficients of other components were of relatively minor importance and all other metal and slag species were also assigned activity coefficient values of one except for Si, P and S in the metal and MnO and P205 in the slag. The coefficients for these species were varied by trial and error until the distribution of these elements between phases closely resembled that found between metal and slag in the iron blast furnace. Values of 0.01, 0.06 and 0.00003 for Si, P and S in the hot metal and 0.4 and 0.05, for MnO and P205 in the slag respectively, were assigned. The activity coefficient values were assumed to be constant and independent of temperature and composition. The value for Si in the metal is about an order of magnitude greater than the value of the activity coefficient of silicon in iron at infinite dilution at 1600°C. Since the reduction of silica from slag is very slow and silicon distribution between slag and metal is usually not achieved in practice the value used in this work is reasonable. The values for P and S in the metal and for P205 in the slag have no real meaning because phosphorus was permitted to occur in the slag as both Ca3(PO4) 2 and P205 and sulphur was permitted to occur in the slag as C.aS with an activity coefficient of one. In all cases, 1 kg of dust was smelted. The objective was to determine the fuel and reductant consumption for a thermally balanced system in which the equilibrium FeO content of slag was 0.1 =t: 0.05% and the carbon content of hot metal was 4.0 + 0.3 %. The smelter was operated at 1450°C.

988

S, M. HAYand W. J. RANKIN

Single Stage Smelting In this model, dusts and flux were smelted in a reduction zone with carbon, and heat was generated by the combustion of methane with oxygen in a separate combustion zone as shown schematically in Figure 1. 100 % stoichiometric oxygen was used (i.e., just sufficient to convert the methane to H20 and CO2). In practice, the complete separation of zones of different oxygen potential may be difficult to achieve and therefore the effect of mixing of zones was examined. Mixing of reaction zones was achieved by dividing the combustion off-gas into two fractions, one of which was equilibrated with the reduction zone while the remainder left the smelter. A fully mixed system occurs when all combustion products are equilibrated with the reduction zone. Zones are fully separated when the combustion products leave the bath without passing through the reduction zone. The heat balances for the two zones were combined to obtain the nett heat balance for the smelter. For particular degrees of inter-zonal mixing, the amounts of fuel and reductant were changed iteratively until the target metal and slag compositions were obtained and the smelter was thermally balanced. Heat losses from the reactor were not taken into account.

Smelter off-gas

Dust and

v

Flux Carbon

Hot metal Slag

Combustion off-gas ~]~iil~i~i[ ~ "~ ir'ii• ' "i ]' " ~",~ i i~rii . [[,,.~i 1'rv " iiff' "iii~il ,!H "~ii~..I]iii:qi! m,~ii!~i[i~ ! iii~ !i " ' "

Biliili~!~ii) ~l]llml[~]iiii

Oxygen b...

............................

, ..................

. ..............

a

Fig. 1 Schematic representation of single-stage smelter model with reduction and combustion zone. Figure 2 shows the variation of carbon and methane consumption as a function of the degree of inter-zonal mixing for the smelting of 1 kg of BOF dust (585 g Fe). The most efficient operation occurred with fully separated reaction zones; as inter-zonal mixing increased from 0 to 100%, methane and carbon consumption increased from 95 and 172 g to 628 and 1585 g, respectively. This trend is expected because the channelling of combustion off-gases to the reduction zone decreases the efficiency of the reduction zone due to the highly endothermic reaction of water vapour and CO 2 with carbon: H20(g ) + C = H2(g ) + CO(g)

(1)

CO2(g ) + C

(2)

=

2 CO(g)

This has the effect of increasing the carbon consumption and places an additional heat load on the process. For a particular degree of inter-zonal mixing, additional methane resulted in more H20 and CO x being directed to the reduction zone and, consequently, more carbon and energy was consumed and more methane was required. At 100% mixing, all the combustion off-gas was converted to CO and H 2, producing smelter off-gas of high calorific value and reducing potential.

Recoveryof iron and zinc from furnace dusts

989

Typical metal, slag and gas compositions are given in Table 2 which shows that 617 g of metal and 262 g of slag were produced with 97 to 99 % of zinc reporting to the gas phase as zinc vapour.

700,

,2000

600 1500

500 0 0

v (D

0

400 ¢o (D tO

lOOO Q

300

(D "s

200

500

100,

0 0

20

4-0

60

80

0 1 O0

I n t e r - z o n o l mixing (~)

Fig.2 The effect of degree of interzonal mixing on methane (fuel) and carbon (reductant) consumption for single stage smelting of 1 kg BOF dusts (100% combustion stoichiometry).

Smelting with Pre-Reduction As a means of recovering some of the calorific value and reducing potential of the smelter off-gas, prereduction and post-combustion steps were incorporated into the model. Figure 3 illustrates a two-zone smelter with a pre-reduction step. In this model, dust is pre-reduced and preheated by the smelter offgas. The pre-reduction step was carried out at 1000°C to ensure that ZnO was reduced and the zinc volatilised in the first stage (boiling point of Zn: 907°C). At lower temperatures, some or all ZnO survived pre-reduction and entered the smelter with the solid feed where it was reduced and volatilised. This zinc reported to the pre-reduction stage where it was re-oxidised and condensed. Under such conditions in practice, zinc would accumulate in the system. In the pre-reduction stage, iron oxides were reduced only as far as FeO for three reasons: metallisation of fine, oxidic, ferruginous materials can create "sticking" problems in practice [2]; it is technically difficult to operate with high levels of post-combustion and pre-reduction [3]; and reduction of FeO to Fe is endothermie and energy savings are possible if this final reduction step is carried out in the smelter. Reduction of the BOF dust at 1000°C to metallic iron requires 1809 kJ/kg dust. Reduction to wustite requires only 159 kJ/kg dust due to the enthalpy gain in oxidising the metallic iron fraction and because 51% of the iron was already present as FeO. Less energy consumed in the pre-reduction stage allows high temperatures to be attained more easily and this is vital for zinc elimination during pre-reduction. As a precaution against metallic iron formation during pre-reduction, the carbon was added to the smelter stage. Beyond 37.5 % inter-zonal mixing in the smelting stage, the reducing potential of the smelter off-gas was sufficiently high to reduce FeO to Fe. The only method of avoiding iron metallisation in the pre-reduction stage whilst maintaining the heat balance at these higher levels of inter-zonal mixing was to use only part of the off-gas for pre-reduction. In these cases, oxygen was added to partially combust the gas before

990

S.M. HAY and W. J. RANKIN

pre-reduction. This raised the oxygen potential of the resulting gas to such a level that metallisation was no longer possible, and additional heat was generated to compensate for the lost heating value of the unused smelter off-gas fraction. In all cases, ZnO, Na20 and K20 were completely reduced to Zn, Na and K and volatilised during pre-reduction and it was possible to balance the two reaction stages thermally.

TABLE 2 Typical equilibrium metal, slag and gas compositions for the smelting of I kg of BOF dust

Species Zn C "Fe

Metal Sla,q Weight (g) Species

Gas Weight (g) Species

0.29

ZnO

0.00

Zn

25.22

FeO

0.50

CH4

Fe203

0.00

02

585.5

Weight (g) 13.5

Si

1.58

SiO2

90.9

SiO

0.04

Mn

3.68

MnO

14.2

Mn

0.07

Ca

CaO

109.8

H2

Variable

AI

A1203

H20

Variable

Mg

MgO

CO

Variable

Ti

TiO2

2.00

CO2

Variable

1.80 39.1

Na

0.07

Na20

0.00

Na

2.23

K

0.04

K20

0.00

K

1.87

P

0.40

P205

0.00

S

0.10

CaS

0.90

Ca3(PO4)2

2.78

Total

616.9

262.0

With less than 37.5 % inter-zonal mixing, the smelter off-gas possessed insufficient sensible heat to supply the energy required in the pre-reduction stage and additional heat was generated by partially combusting the off-gas. The formation of FeO and the removal of Zn, Na and K during pre-reduction were not affected by this step. The extra processing involved in separating the smelter off-gas into two fractions and adding oxygen to the pre-reduction stage represents an additional cost though the unused fraction of the smelter off-gas has a high calorific value, especially at high levels of inter-zonal mixing, and may be of some use in practice.

Recoveryof iron and zinc fromfurnacedusts

991

Process off-gases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Dust and ~ Oxygen

.............. ! ........ t

.........................

off-gas

i.iotm ... i _ l . . . . . . . . . . . . . . . . . .

Carbon

-~

Slag

Combustion

off-gas [

Methane Oxygen

lI '1

-

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

S M E L T E R STAGE Fig.3 Schematic representation of model for smelting with pre-reduction. Figure 4 compares carbon consumption rates for pre-reduction-smelting with single-stage smelting. Significant fuel and reductant savings are evident with pre-reduction, especially at high inter-zonal mixing rates in the smelter. Pre-reduction lowers the amount of carbon required for reduction reactions in the smelter and this lessens the effect of Eqs. 1 and 2 and results in lower methane consumption. As in the case of single-stage smelting, this process operates most efficiently at low degrees of inter-zonal mixing.

Smelting with Post-Combustion A post-combustion stage was added to the two-zone smelter model, described previously, as shown in Figure 5. As in the previous cases, the mixing of combustion products with the reduction zone was varied. In this case, however, the off-gas from the reduction zone was combined with unused combustion off-gas and this gas was partially combusted with oxygen in a post-combustion zone. A fixed proportion of the heat generated by post combustion was transferred to the reduction zone; this is referred to as the Heat Transfer Efficiency (HTE) of the process. Heat not transferred left the system as sensible heat in the post-combustion off-gas. The heat balances for the reduction and combustion zones were combined to obtain the nett heat balance for the smelter stage. HTE values of 60, 75 and 90% were examined. The smelter off-gas was post-combusted to an oxidation degree of 50 + 1.0% calculated as follows: %OD

---

(%C0: + %H20) x ]00 (%CO2 + %H20 + %CO +%H2)

(3)

992

S.M. HAY and W. J. RANKIN

The resultant ratio gives an indication of the extent of combustion; gases with an oxidation degree of 100% are fully combusted. Oxygen entered the post-combustion stage at 25°C. All other parameters were as described previously. For particular degrees of inter-zonal mixing and HTE, the amounts of fuel, reductant and oxygen were varied iteratively until the level of carbon in metal, F e e in slag and the oxidation degree of the post-combustion off-gas fell inside the target range and until the system was thermally balanced. 1600

I

1400

I

I

I

0 Smelting Only

• Smelting with P r e - r e d u c U o n 1200

"S C

o

1000 800

o

600 400

0'

0

20

40

60

80

1O0

Inter-zonal mixing (~)

Fig.4 Comparison of the effect of degree of interzonal mixing on carbon consumption for smelting 1 kg BOF dust with pre-reduction. The effect of post-combustion and HTE on carbon and methane consumption is shown in Figures 6 and 7. Post-combustion of the smelter off-gas was not required when inter-zonal mixing was less than 25 % because the oxidation degree of the smelter off-gas was greater than 50%. From 25 to 100% inter-zonal mixing, the oxidation degree of the smelter off-gas decreased from 50% to about 1% and, consequently, an increasing amount of heat was generated by post--combusting this gas to an oxidation degree of 50 %. Therefore, as the amount of inter-zonal mixing was increased, an increasing fraction of the heat required by the smelter was supplied by post-combustion. This is shown in Table 3. As a result, methane consumption decreased as the degree of inter-zonal mixing was increased; the greater the HTE value, the greater was the decrease. Post-combustion does not introduce oxidised species into the reduction zone and as a result, carbon consumption does not increase with mixing as dramatically as for single-stage smelting (Figure 2). The sensible heat of the post-combustion off-gas leaving the system represents a considerable heat loss. The loss is greatest when low HTE is combined with high levels of inter-zonal mixing (Table 3) and operation of the process within this range would be undesirable due to the heat loss and difficulty in handling gases at these temperatures. Maximum efficiency for single-stage smelting and for smelting with pre-reduction occurred with fully separated reduction and combustion zones. With post-combustion, maximum efficiency with respect to methane consumption occurred at high levels of inter-zonal mixing, and carbon consumption was much lower than in the other two cases. This is an important finding as it shows that operation of this process with any degree of inter-zonal mixing could be feasible.

Recoveryof iron and zinc from furnace dusts

993

Oxygen

Dust and flux

Hot metal Slag

Carbon Combustion off-gas Methane Oxygen

i | i

. . . . . . . . . . . .

. . . . . . . . . . . . . .

. o . o . . . . . o

N

. . . . . . . . . . . . . .

. . . . o . . o . . . . . .

Fig.5 Schematic representation of model for smelting with post-combustion.

400

I

I

|

I

0 6 0 ~ HYE • 75=: HTE [] 9 0 Z HT~

350

./.,(

~ J

i

,

-

J

300 250 20O 150

0

,

20

,

t

,

40 60 80 Inter-zonal mixing (~)

1O0

Fig.6 The effect of inter-zonal mixing and heat transfer efficiency on carbon consumption for smelting of 1 kg BOF dust with post-combustion.

994

S.M. HAy and W. J. RXNKIN

120

I

I

I

I

20

4O

60

80

100

ua

o

80

.c:

6O

4O 0

I O0

Inter-zonol mixing (~)

Fig.7 The effect of inter-zonal mixing and heat transfer efficiency on methane consumption for smelting of 1 kg BOF dust with post-combustion. TABLE 3 Off-gas Temperatures and Percent of Total Smelter Energy Supplied by Post-Combustion. Inter-zonal Mixing (%)

75% HTE PCOGT FTEPC (°C) (%)

90% HTE PCOGT FTEPC (°C) (%)

0

1450

0.0

1450

0.0

1450

0.0

12.5

1450

0.0

1450

0.0

1450

0.0

25

1683

12.1

1599

16.1

1514

19.8

37.5

1857

22.3

1716

28.0

1563

33.4

50

2024

30.2

1824

37.0

1605

42.4

62.5

2164

36.7

1901

43.8

1645

51.4

75

2274

42.1

1979

49.3

1668

57.9

87.5

2363

46.7

2048

54.1

1686

60.6

2434

50.5

2103

58.4

1712

64.2

100

Legend:

60% HTE PCOGT FTEPC (°C) (%)

PCOGT

=

FTEPC

=

Post-combustion off-gas temperature Fraction of total smelter energy supplied by post-combustion

Recovery

of iron and zinc from furnace dusts

995

Smelting with Post-Combustion and Pre-Reduction Aspects of the separate post-combustion and pre-reduction models were combined to develop the model shown schematically in Figure 8. The combined off-gas from the combustion and reduction zones was post-combusted to an oxidation degree of 50 ± 1.0 % and 90 % of the heat generated by post-combustion was transferred to the reduction zone. Incoming BOF dust was pre-heated to 1000°C and all iron species in the dust were converted to FeO in the pre-reduction stage by the off-gas from the post-combustion zone. The three-zone smelter and the pre-reduction stage were thermally balanced separately.

................................. t .......... iiillilllllllllllll!lillillill~lli =,~';.='~'H=.~, i lllIIIIl!llllE~lll$11il| II~l l i i l i | l l l l l ~ l ~ l l l

Dust a n d flux

~l~'q

• ~t

i'II

i~I!I~=iIIIIIIiI,ilII= iE°°~I' ~="~"i .....IIlI IiiIII u~=I il~lillll¥11ti! ~"

IilIIII$I$IIIIIIIII t i i i

i= I!=II li ~ A I ! i

- ............

.---.

.......

$IIIIIIiiliiiiiiiiiiiii

II i i ! i

. . . . . . . . . . . . . . . . . . . . . .

...d

Pre-reduced dust and flux

Smelter off-gas > 1450 ° C ; i

Oxygen

I

: i

ii lillHlI~III$111115IiI$IIIIIIilIiili!£ilIiil~l~t~H~¢~N ~ ' jili#i!' "~ = ' =°.......... " ", =.,t=. !. ,i~ililili!iiliiii I

-"N==I

!!!iliillH iiiiiii!Ilil~:,

l~iili!I hlillli

: .III IIII,,....=

Reduction and | Combustion / off-gases

Heat

,=,~o,.I.,~.... :

~II~.......IiliiliiiiI Iii I~

i,l,~lil,l l i l i lI

'i'

" " ~

'.l ~!

.

li

II f I

H o t metal

I i

slag

I ~-ll=,l.=-h=~ .... = ~ l l ' ' " i=i

=i

i :!IIrI L,I I l i i i lI~i~iliII I ili~iiI~i~iI £l~!iliiiili!iii!l!i!i!lI

Methane Oxygen '. . . . . . .

..

.......

.oo.

......

SMELTER

.°

........

° .............

. ..........

•

STAGE

Fig.8 Schematic representation of model for smelting with pre-reduction and post-combustion. The post-combustion off-gas temperatures obtained were very similar to those in Table 3 for 90 % HTE. The sensible heat of these gases was insufficient to pre-heat dust to 1000°C and, therefore, oxygen was added to generate additional heat by combustion. Pre-reduction of the dust was uncomplicated; metallic iron did not form because the oxidation degree of the reducing gas was always about 50 %.

996

S.M. HAYand W. J. RANKIN

Comparison of the Smelting Processes Figures 9 and 10 compare carbon and methane consumption versus degree of inter-zonal mixing for the four process options. Over the full range of mixing, smelting with post-combustion and pre-reduction is clearly the most efficient. This is because parts of both the chemical energy and sensible heat of the smelter off-gas is usefully used in the process. Although this process is the most efficient, in terms of removal of oxygen from the dust, the pre-reduction step is relatively ineffective; only 11.5 % of the oxygen originally combined with iron oxides is removed. This is because much of the iron in the dusts is already present as FeO and because the metallic iron initially in the dust was oxidised to FeO. The value of the preheating step, however, cannot be denied. 1600

I

o ~u~

140o

I

I

~y

• Smeltln8 with pre-reducUon

/

t

0 SmelU~i with post-oombusUon / V Smelting with pro-reduction / and p o ~ - o o m b u a U ~ /

12oo i

I

lOOO

600 8

400

200, 0

I

0

I

20

I

I

40 60 Inter-zonal mixlng (~)

1O0

80

Fig.9 Comparison of the effect of inter-zonal mixing on carbon consumption for smelting 1 kg BOF dust for the four smelting options (for post-combustion, an Oxidation Degree of 50 % and Heat Transfer Efficiency of 90 % were used).

700

I

I

I

I

o amea~i ox~y

600

A Smolttui phm pre-reduoUon [] 9melUn4iplua polt-oombumtion V Sm~Unl ph= I n - . n o t i o n and Imat-

5O0

/

/

i°

3O0 2OO 100 0

i

I

I

I

20

40

60

80

1O0

Inter-zonal mixing(~) Fig. 10 Comparison of the effect of inter-zonal mixing on methane consumption for smelting 1 kg BOF dust for the four smelting options (for post-combustion an Oxidation Degree of 50% and Heat Transfer Efficiency of 90% were used).

Recoveryof iron and zinc from furnace dusts

997

Smelting of BOF and Blast Furnace Dust Simultaneously Blast furnace dust typically comprises about 25 % of all dust generated in a steelworks. A dust treatment process should be able to treat both dusts, preferably simultaneously. Therefore the models previously described were used to examine the treatment of a dust blend containing 75 % BOF dust and 25 % BF dust. Smelting of 1 kg of such a blend will yield 7.5% less iron than smelting BOF dust alone because of the lower iron content of the BF dust. The carbon content of BF dust will aid reduction reactions in the smelting vessel. Smelting of the blend with pest-combustion of off-gases gave almost identical results to similar treatment of BOF dust although the consumption of carbon was slightly lower. Pre-reduction of the dust blend followed by smelting was modelled by two different methods. In one case, only the BOF dust fraction was pre-redueed and the blast furnace dust was added at 25°C to the smelter stage to prevent metallisation of iron (by carbon in the BF dust) in the pre-reduction stage. All other factors were the same as before. Table 4 compares selected results for this case with the results for the same treatment of BOF dust. More methane was required for smelting the blend because one quarter of the feed entered at 25°C. Carbon in the blend led to lower carbon consumption at low and intermediate degrees of inter-zonal mixing. At high levels of inter-zonal mixing this benefit was offset by the greater amount of oxidising gases entering the reduction zone. T A B L E 4 Carbon and Methane Consumption for the Smelting with Pre-reduction for BOF Dust and BF/BOF Dust Blend.

BOF dust Inter-zonal mixing (%)

Dust blend

Partial pre-reduction

Partial pre-reduction of BOF dust

C (g)

OH4 (g)

C (g)

OH4 (g)

0

153

62

100

67

50

274

108

232

118

100

1055

401

1100

445

Total pre-reduction of both dusts C (g)

CH 4 (g)

207

80

r

In the second case examined, the entire blend was pre-heated and pre-reduced by smelter off-gas; complete metallisation of the iron content was allowed to occur. The reduced product was then melted in the two-zone smelter stage. This process was feasible only with fully mixed zones; with less than full mixing, the smelter gases had insufficient sensible heat and reducing potential to fully pre-heat and prereduce the dust. Though full metallisation was not crucial, a temperature of 1000°C was essential to ensure complete zinc removal in the pre-reduetion stage. When the smelter zones were fully mixed, carbon and methane consumption was much lower than for cases where the BOF dust fraction was partially reduced, as shown in Table 4.

Smelting of Dusts with Coal and Air The BOF and BF dust blend was smelted using coal and air rather than carbon and oxygen as used in the previous calculations in order to compare the process concept with an iron blast furnace operation. The model for smelting with pest-combustion was used. Gases were post-combusted to an oxidation degree of 50 + 1.0%, inter-zonal mixing in the smelter was maintained at 100%, and an HTE value of 90% was imposed. Coal ash was assumed to be 10% by mass of the coal and to consist of 8.4% Fe203, 60.4% AI203, 30.2% SiO 2 and 1.0% CaO. The volatile matter was assumed to consist of 37.4% CH 4, 14.1% 02, 3.1% N2and 1.0% H2bY mass.

998

S.M. HAYand W. J. RANKIN

Two coals containing 10 and 30% volatile matter, respectively, were used as reductant. Smelting 1 kg of the dust blend to metallic iron consumed 400 g of coal and 94 g of methane, for the low volatile coal, and 430 g of coal and 78 g of methane, for the high volatile coal. These figures are equivalent to a coal consumption rate of 685 and 735 kg, respectively, per tonne o f hot metal. This compares with iron blast furnace operation for which typical coke consumption rates are 440 - 480 kg per tonne of hot metal. The methane consumption was equivalent to 161 kg per tonne of hot metal which is significantly higher than consumed in modern blast furnaces, typically around 30 to 50 kg of natural gas per tonne of hot metal.

SOLID STATE R E D U C T I O N The technology for producing zinc-free metallised and non-metaUised clinkers from steelworks dusts at temperatures around 1000°C is well known. Waelz kilns, in which the degree of iron oxide reduction can be controlled by varying reaction conditions, are currently in use producing both types of clinkers, primarily from EAF dusts. In this process, zinc is reduced and volatilised using solid and gaseous reductants. The treatment of BOF and BF dusts under similar conditions was modelled so that the predicted fuel and reductant consumption rates could be compared with the previous cases. A single stage was employed and heat was supplied by the combustion of methane and the iron content of dusts was partially or fully reduced by carbon or gaseous products resulting from the partial combustion of methane. When BOF dust was heated from 25 to 1000°C by partial combustion of methane (50 to 80% stoichiometric oxygen), no carbon was needed to reduce ZnO; CO and H 2 resulting from the combustion reactions acted as the reductants. Minimum methane consumption and maximum zinc concentration in the off-gas occurred at 62.5% stoichiometric oxygen. Zinc, sodium and potassium were completely volatilised and all iron species were converted to FeO. Because of the carbon content of BF dust, metallisation of the iron fraction occurred in some cases when the dust blend was examined. Na20 and K20 were reduced and volatilised. Table 5 lists the reagent consumption and major element distributions for the various situations examined. All reagent weights listed are the minimum amounts required to thermally balance the system and remove zinc and, in some cases to metallise the iron. The table shows that it was possible to fully or partially reduce the dust blend with carbon and/or methane by varying the amount of oxygen. Reagent consumption was greatest for cases involving iron metallisation. T A B L E 5 Reagent Consumption and Final Form of Iron for the Removal of Zinc from 1 kg of Dust Blend at 1000°C.

Final form of iron

Reagent consumption Carbon (g)

Oxygen (g)

Methane (g)

27

132

0

FeO

207

290

0

Fe

0

338

180

Fe

1'22

312

78

Fe

0

160

40

FeO

Recovery of iron and zinc from furnace dusts

999

DISCUSSION Table 6 gives reagent consumptions for the most efficient smelting and solid-state processes. For the single-stage smelting of BOF dust and for smelting of partially reduced and preheated BOF dust in the two-stage pre-reduction process, maximum efficiency occurred when reduction and combustion zones in the smelter were fully separated. Very different oxygen potentials exist in the zones and inter-zonal mixing results in the transfer of oxygen from the oxidising zone to the reducing zone. Oxygen transfer in slag is promoted by bulk mixing and by the FeO/Fe203 redox reaction and, therefore, a low total iron content of the slag is important with respect to zonal separation. TABLE 6 Comparison of Reagent Consumption for the Various Processes Modelled

Dust

Thermodynamic model

Inter-zonal mixing (%)

Temp. (°C)

Final form of iron

Reagent Consumption (g) C

CH4

02

BOF

SM + PC + PR

100

1450

Fe(I)

210

25

100

BOF

SM + PC

100

1450

Fe(I)

284

50

200

0

1450

Fe(I)

172

97

388

153

62

248

BOF

• SM

BOF

SM + PR

0

1450

Fe(I)

BOF

Low Temp. DZ

-

1000

FeO(s)

0

53

133

Blend

Low Temp. DZ

1000

FeO(s)

0

40

160

Blend

Low Temp. DZ

1000

FeO(s)

27

0

132

Blend

Low Temp. DZ

-

1000

Fe(s)

207

0

290

Blend

Low Temp. DZ

-

1000

Fe(s)

122

78

312

Blend

Low Temp. DZ

-

1000

Fe(s)

0

180

338

Legend:

SM

= smelting

PC

= post-combustion

PR

= pre-reduction

DZ

= dezincification

Blend = 75% BOF dust + 25% blast fumace dust

1000

S.M. HAYand W. J. RANKIN

In the DIOS process, there are two zones within a deep slag bath. Oxygen is injected into the top zone to post-combust reduction gases and reduction occurs in the bottom zone by reaction of high-carbon iron droplets (ejected from the hot metal by gas injected into the metal) with FeO in the slag. Zones with different oxygen potentials can be maintained in relatively shallow Sirosmelt slags by controlling injection and addition conditions though the factors that affect the degree of inter-zonal mixing are not well understood. This selectivity means that a Sirosmelt-type reactor might be suitable for single-stage smelting of dusts. In other kinds of reactors, zones could be separated by, for example, injecting dusts and reduetant into the hot metal through tuyeres and combusting fuel in the slag. In such a process, iron oxides would be reduced in the carbon-containing iron bath and heat would be transferred from the slag to the metal. With the reduction zone lying below the combustion zone, oxidising combustion off-gases would leave the bath without mixing with the reduction zone. Zonal separation would be enhanced in all these processes by a low level of FeO in the slag. With high levels of inter-zonal mixing in the smelter, the most efficient operation occurred when smelting was combined with high levels of post-combustion. This appears to be the regime of operation of the Hlsmelt process. A high degree of heat transfer from the post-combustion zone to the bath is necessary to limit the post-combustion off-gas temperature to a practical level. Post-combustion rates of around 50% have been achieved in the Hismelt process simultaneously with high levels of heat transfer. In all the smelting cases examined, all the zinc reported to the gas phase. If, in practice, dust carry-over could be minimised, the fume extracted from the process gases would be predominantly ZnO formed by reoxidation of the zinc vapour. This by-product may be a suitable feed for a zinc producer depending on tolerance levels for PbO, Na20 and K20. All sodium and potassium were removed from the iron product. The suitability of the molten iron for treating in the BOF, therefore, probably depends on the levels of sulphur and phosphorous in the metal. Little information on the likely deportment of these elements could be ascertained from the thermodynamic study because their distribution between phases depends on the composition of the slag and a more detailed thermodynamic and experimental investigation would be necessary. The final slag contained no hazardous species (see Table 2) and, in practice, a slag of this composition, could be safely used or dumped. Thus, the three products resulting from the smelting of dusts, the ZnOrich fume, hot metal and slag, could probably be further processed or safely disposed of and the criteria stipulated by Piret and Miiller [1] for environmentally safe, if not economically profitable, treatment of dusts are satisfied. Results for the reduction of dusts at 1000°C indicated that reagent consumption rates were lower when dezincing dusts at this temperature compared to most cases in which dusts were smelted though it should be noted that oxygen was used so the results could be compared with smelting. In practice, air would be used and reagent consumption would be higher because of additional energy required to heat the nitrogen. In these relatively low temperature processes, the minimum consumption of fuel and reductant occurred when no metallic iron was formed. The FeO-rich clinker produced from such a process would probably be a suitable additive to the BOF, replacing iron ore as coolant. The use of clinker in this manner would depend on the residual concentration of phosphorous, sulphur and other deleterious species. All the phosphorous in the dust reported to the clinker and, although all the sulphur reported to the gas phase, incomplete sulphur removal from the clinker is likely in practice due to the formation of compounds such as FeS and CaS. The high level of CaO in the clinker could be of benefit to BOF operation as it reduces the amount of flux required. An experimental study would be necessary to confirm the behaviour of sulphur and phosphorus. In most steelplants, it is likely that the BOFs would not be able to consume the total amount of clinker produced if all dust generated on site was treated. Addition of excess clinker to the blast furnace would be an option if there was excess capacity in the coke ovens and sinter plant. The BOF may be able to consume a greater amount of clinker if it was fully metallised. In this case the clinker would act as a replacement for scrap steel. An advantage of a smelting process over a lower temperature process is that

Recoveryof iron and zinc from furnacedusts

1001

because molten iron and slag are produced, some degree of phosphorus and sulphur removal from the metal is possible, whereas with clinker-producing processes less sulphur and little, if any, phosphorous would be removed from the product. This would put a greater burden on the hot metal treatment steps. A smelting process would have to be used if the production of steel was limited by the capacity of the coke ovens or sinter plant rather than the BOFs as it is a parallel process, producing hot metal, which does not require coke.

CONCLUSIONS From the thermodynamic analysis undertaken, it can be concluded that the production of metallic iron and volatilisation of zinc from BOF and BF dusts is feasible under a variety of conditions. There were few differences between hot metal and metallised clinker-producing processes in terms of the removal of zinc, sodium and potassium, and potential end uses of the iron-containing product. For smelting processes to be competitive with existing clinker-producing processes the most efficient smelting regimes must be employed. This means separation of the combustion and reduction zones in the smelter or use of prereduction and post-combustion stages, with a high level of post-combustion heat transfer efficiency, is essential. An advantage of smelting dusts over producing clinker is the possibility of removing some of the sulphur and phosphorous in the slag. Hot metal produced by smelting could be charged to the BOF and, thus, this option would place no additional load on the coke ovens, sinter plant or blast furnace. Metallised clinker could be used to replace scrap steel in the BOF. Non-metallised clinker could be used to replace iron ore as coolant in the BOF and/or be recycled to the blast furnace. The preferred option for treating BOF and BF dusts will depend strongly on local circumstances.

ACKNOWLEDGMENTS Financial assistance for this work, including a Postgraduate Studentship for SMH, was provided by the Australian Mineral Industries Research Association Limited. The assistance of Broken Hill Proprietary Limited in providing technical advice is gratefully acknowledged.

REFERENCES .

.

3. 4.

.

ME 7/8--D

Piret, N.L., & Mfiller, D. Criteria for the selection of a recycling process for low zinc containing residuals from the iron/steel industry. Prec. Int. Syrup. on Processing of Residues and Effluents, San Diego, March, 1992, TMS (1992). Wright, J.K., Taylor, I.F. & Philp, D.K. A review of progress of the development of new ironmaking technologies. Minerals Engineering '91, Singapore, (Feb. 1991). Wright, J.K., Direct smelting of iron ore. p.ll.1 Prec. Mervyn Willis Symposium, (Eds. M. Nilmani and W.J. Rankin) University of Melbourne (1992). Rankin, W.J. et al. Process engineering of Sirosmelt reactors: lance and bath mixing characteristics. Extractive Metallurgy '89, Institution of Mining and Metallurgy. London, 577 (1989). Bale, C.W. & Eriksson, G. Metallurgical thermochemical databases - a review. Canad. Metall. Quart., 29, 105-132 (1990).