Development of fluxed blast furnacepellets with application of coatings

Lawrence Hooey1

Mats Hallin1

Kalevi Raipala2

1) LKAB R&D Metallurgy, Box 952 SE-971 28Luleå, Sweden

2) Rautaruukki Corporate R&D, Fundia Koverhar,Lappohja, FIN-10820, Finland

SUMMARYHigh iron content fluxed pellets have been tested inpilot and full-scale trials. The reduction behaviour ofthe pellets in both pilot-scale and commercial blastfurnaces was acceptable. The furnaces' behaviourwith the experimental pellets was acceptable and hadthe potential to improve the blast furnace operations.However, there was an unexpected result in the full-scale trials: the hot metal carbon content was lowerand sulphur distribution poorer than when usingregular olivine pellets. This behaviour appears to beassociated with high temperature clustering andmeltdown properties of the fluxed pellets that havebeen observed in dissections of the experimentalblast furnace. Further testing in the experimentalfurnace showed that application of either quartzite orolivine coating at 3.6 kg/t pellet restored hot metalquality and shows potential for improving blastfurnace stability and reducing problems associatedwith alkali circulation.

1. INTRODUCTIONIn 1997 LKAB commissioned the Experimental BlastFurnace (EBF) in Luleå, Sweden. The role of this1.2m hearth diameter pilot blast furnace was toprovide an intermediate step between laboratory andfull-scale testing of experimental pellets. The furnaceis a complete blast furnace operation that producescirca 35 tonnes/day of hot metal and has beendescribed previously.1,2 The EBF has been run in 11campaigns of about 8 weeks each. After extensivetesting in the EBF, a new grade of acid pellets for usewith sinter, the KPBA pellets, were introduced in2001.3

In addition to pellets designed for use with sinter, anew grade of fluxed pellets designed for the Nordicblast furnaces of SSAB and Fundia Wire are beingdeveloped. The furnaces are currently burdened withLKAB's olivine pellets (MPBO and KPBO).

In order to modify the properties of the pellets to suitthe furnace operation and long term developmentplans, the pellet chemistry and reduction behaviourmust be considered. Nordic blast furnaces have anumber of features that have acted as the drivingforce for pellet development. Among the main

features are desire for very low slag volumes (150kg/tHM or even lower), high productivity (circa 3t/d/m3 w.v.) and high rates of injectants and oxygen.With these objectives in mind, the suitable pelletproperties are listed in Table 1.

The factors listed in Table 1 are the mainconsiderations, of course provided mechanicalstrength, pelletising properties, particle size and soon, are all satisfactory. If the pellets can be produced,the factors in Table 1 are all affected by the choice ofadditives.

Tab. 1. Summary of desired pellet behaviour andfurnace operation.

PelletProperties

Furnace Operation

PelletChemistry

Suitable for flux injection and100% pellet operation

Slag rate reduction

High oxygen injection (low topgas temp.)

HighReductionstrength, Lowswelling

High productivity, stableoperation

Reducibility Low residence timesHightemperatureproperties

Low cohesive zone, stableburden descent

Meltdown andslag formation

Dripping behaviour of slag andiron meltdown must be goodfor very low slag volumes

2. DEVELOPMENT OF FLUXED PELLETS WITHCOATINGS

With the objectives listed in Table 1, development ofthe fluxed pellets followed the scheme in Figure 1.Various pellet types were tested in laboratory and theEBF before being introduced to industrial furnaces.

The EBF is quenched at the end of each campaign inorder to evaluate the material behaviour in thefurnace. Nitrogen is introduced in the top andremoved via the tuyeres to prevent oxidation and aheat-front from moving up the furnace. The furnace isthen excavated layer by layer with extensivesampling and evaluation of materials.

In campaign 4 excavation it was noted that the pelletshad very high reduction strength - that is the pelletshape was maintained quite low in the furnace andthere was no indication of swelling or cracking ofpellets. A detailed comparison with other pellets isbeyond the scope of this paper.

Pellet production in full scale

Pellet testing in full scale

FINALPRODUCT

Pellet production in lab scale

Pellet testing in lab scale

Pellet production in pilot scale

Pellet testing in pilot scale

EBF

Fig. 1: LKAB's Pellet development path

One observation that was noted, however, was atendency for pellets to form clusters. Thephenomenon appears to be a solid-state sintering ofiron during reduction. Figures 2a and b showsclusters forming in the middle shaft of the furnace inthe campaign 4 excavation. The clustering, however,did not affect the furnace stability and was notreflected in the furnace operation.

It is well established that alkalis can have negativeimpact on reduction behaviour in the blast furnace -such as swelling4. It was decided to test coating ofthe pellets to prevent sticking, and especially in thecase of the fluxed pellets, to see if alkali absorptioncould be improved.

Fig. 2a: Clustered pellets forming in layer 12 ofexcavation of campaign 4 with proto-type MPB1pellets. The height above tuyeres was circa 2.4 m.Degree of metallisation was circa 36%at mid-radius.

Fig 2b: Close-up of clustered pellets in layer 15 (3pellet layers below Fig. 2a. The height above tuyerelevel was 2.0 m. Degree of metallisation was 71% atmid-radius.

For these general reasons, in EBF campaigns 7 and8 fluxed pellets were tested with applications ofolivine, quartzite and dolomite sprayed in slurry formonto the pellets. The coating amount was chosen at3.6 kg coating material/tonne pellets plus 0.4 kgbentonite to improve the binding of the coating.

3. Laboratory Evaluation of Fluxed Pellets

After various experimental pellet types had beentested, a type called 'MPB1' fluxed pellet emerged ashaving properties and chemistry that were the mostsuitable for possible replacement of MPBO pellets inNordic blast furnaces.

The MPB1 fluxed pellet composition and metallurgicalresults are compared to MPBO in Tables 2 and 3.The chemistry of the MPB1 is suitable because itallows removal of some of the limestone charged tothe furnace. This in turn decreases the requiredthermal energy and coke rate. The metallurgicalproperties as measured in the laboratory are as goodas or superior to MPBO with higher compressionstrength, LTB, higher softening temperature andlower pressure drop.

Tab. 2: Chemistry of MPBO and MPB1 Experimentalpellets

MPBOOlivine Pellets

MPB1ExperimentalFluxed Pellets

Fe 66.8 66.5CaO 0.29 1.65MgO 1.46 0.41SiO2 2.15 1.70Al2O3 0.41 0.38CaO/SiO2 0.13 0.97

Tab. 3: Metallurgical properties of MBPO andMPB1 pellets

MPBO MPB1Cold Strength (daN) 213 289ISO 13930 LTB (%+6.3) 74 80ISO4695 R40 (%/min) 0.54 1.1TRT7992 dP (mmWg) 18 2.4High temperaturesoftening temperature* oC 1263 1310Swelling (LKAB test)** 32 17*Rautaruukki's softening and melting test** LKAB's swelling test using higher temperature(1000oC) that yields higher values than the ISO4698test (900oC)

4. EBF TESTING OF FLUXED PELLETS

The final MPB1 testing was made in the EBF incampaigns 7 and 8. Tables 4,5 and 6 showsummaries of operating data.

As the furnace operating periods are typically 2-3days long the furnace fuel rate cannot necessarily beoptimised. Furnace stability is considered the mostrelevant for comparing pellets. The standarddeviations in ETA CO, burden descent rate andburden resistance index are used to evaluate thestability. Production statistics such as fuel rate andproductivity can be compared for longer experimentaltimes or if extreme behavioural differences arepresent. Table 4 and 5 show that the stability of theMPB1 is the same or better than MPBO. Table 6shows that there were no significant differences infuel rate or production in pilot scale.

Coated MPB1-type pellets were also tested incampaigns 7 and 8. In campaign 7 the goal was toestablish if the coating technique and to see if thecoatings remained on the pellets after coating,transportation, screening and charging to the furnace.This preliminary testing proved successful, but theperiods were too short to achieve a very reliablecomparison. Comparison of flue dust generation inFigure 3 showed that coating material was not beingremoved via top gas. Chemical assays also showedthat no appreciable coating was lost in transport andhandling.

0

2

4

6

8

10

12

MPB1 MPB1 +Dolomite

MPB1 +Olivine

MPB1 +Quartzite

MPBO

kg f

lue

du

st/t

hm

Fig. 3: Flue dust generation measured in campaign 7for coated and uncoated pellets.

In campaign 8, the coated MPB1 pellets were testedfor longer periods (circa 2-3 days) in the EBF. Table 5shows the basic results of stability were againcomparable to MPB1 or MPBO.

Tab. 4: Summay of furnace operation in Campaign 7testing of MPB1 Pellets

-----ETA CO---- ---PV Bosh --- DescentTime

hAverage STD Average STD STD

MPBO 27 45,6 1,0 5,9 0,2 1,2MPB1 76 46,1 0,9 6,3 0,9 0,7Pvbosh =(Pblast

2-Ptop2)/Vbosh1.7 where P is in atm

absolute; Vbosh = bosh gas volume Nm3/s/m2 hearthareaBDR= burden descent rate, cm/min

Tab. 5: Summary of furnace operation in Campaign 8testing of MPB1 pellets, and MPB1 pellets withcoatings of quartz (MPB1-Quartz) and olivine (MPB1-Olivine).

-----ETA CO---- ---PV Bosh --- BDRTime

hAverage STD Average STD STD

MPBO 60 47.6 1.0 6.0 0.3 0.55MPB1 42 47.4 1.1 7.3 0.6 0.52MPB1-Quartz

67 46.9 0.9 7.2 0.7 0.35

MPB1-Olivine

76 47.5 1.4 6.6 0.4 0.48

Tab. 6: Summary of furnace operation in Campaigns7 and 8.

Prod.t/h

BlastNm3/h

BlastO2

%

Cokerate

kg/tHM

Coalrate

kg/tHMCam.7MPBO 1.32 1721 22.5 441 90MPB1 1.32 1725 22.6 442 98

Cam.8MPBO 1.55 1737 24.7 403 127MPB1 1.56 1738 24.7 400 123MPB1-Quartz

1.54 1744 24.7 400 127

MPB1-Olivine

1.57 1744 24.7 396 124

5. FULL-SCALE TESTING OF FLUXED PELLETS

The MPB1 pellets were tested in two industrialfurnaces described in Table 7. The operationsdiffered slightly with Fundia furnace operating with oilinjection and SSAB Oxelösund operating with coalinjection. Both furnaces continued to use other pelletsin the burden during the one to two week trial periods.

The results of the trials showed little change in thefuel rate and production rate statistics.

Tab. 7: Blast furnaces used for full-scale testingFundia SSAB

Oxelösund #2Working volume m3 567 760Productivity t/m3/d 2.9 2.5Injectants kg/tHM Oil c. 90 Coal 95O2 in blast % 26.5 24Slag rate kg/tHM c. 160 c. 155

Tab. 8: Summary of BF production and fuel rate atSSAB Oxelösund

Timedays

Burden Prod.Ratet/d

Fuel rateKg/thm

Ref. 18 70% MPBO30% KPBO

1920 473

Test 8 c. 60% MPB140% KPBO

1920 472

Tab. 9: Summary of BF production and fuel rate atFundia Wire, Koverhar.

Timedays

Burden Prod.Ratet/d

Fuel rateKg/thm

Ref. 14 80% MPBO20% other

1553 466

Test 8 60% MPB120% MPBO20% other

1548 466

However, the behaviour of the silicon, carbon,sulphur and potassium in the slag and hot metalshowed unexpected but very consistent trends inboth the industrial furnaces. Looking at therelationships between silicon and carbon, bothfurnaces showed a drop in carbon content forequivalent hot metal silicon content (Fig. 4 and 5).

The MnO/Mn relationship, a good indicator of theoxygen potential in the hearth, showed that theoxygen potential increased when MPB1 pellets wereused (Tab. 10).

Tab. 10: Relationships showing (MnO)/[Mn] ratiosindicating higher oxygen potential in the high MPB1periods.

ReferencePeriod

High MPBI Period

Fundia Wire 2.39 2.69SSAB Oxel #2 1.62 1.98

The increase in oxygen potential is clearly reflected inthe poorer desulphurisation relative to alkali output.Figures 6 and 7 show that the distribution of sulphurbetween slag and metal became poorer forequivalent alkali output - which means that the

furnaces would have to operate either with higheralkali circulating loads or with higher sulphur contenthot metal. Either way, the behaviour was undesirable.

The reason the oxygen potential increased occurredcannot be determined directly from the trial data.However, it was thought that the clustering behaviourseen in the excavation might have an impact on thefull-scale furnaces that was not visible in the EBF. InEBF campaign 8, which was running in parallel to thefull-scale trial, was evaluated in more detail for thehot metal-slag quality relationships.

3,2

3,6

4,0

4,4

4,8

0,0 0,2 0,4 0,6 0,8 1,0[Si]

[C]

MPBO(>80%) MPB1 (>60%)

MPBO

MPB1

Fig. 4: Fundia Wire results showing a drop in hotmetal carbon content for a given silicon content.

3,8

4,2

4,6

5,0

0,3 0,5 0,7 0,9[Si]

[C]

MPBO 70%, KPBO 30% MPB1 70%, KPBO 30%

Fig. 5: SSAB Oxelösund results showing a similardrop in hot metal carbon content for a given siliconcontent.

0

25

50

75

100

0,2 0,4 0,6 0,8 1,0 1,2K2O Content of Slag

(S)/[S]

MPBO (>80%) MPB1 (>60%)

MPBO

MPB1

Fig. 6: Fundia Wire results showing poorerdesulphurisation for a given alkali output.

0

10

20

30

40

50

0,2 0,4 0,6 0,8 1,0Slag K2O Content

(S)/[S]

MPBO 70%, KPBO 30% MPB1 70%, KPBO 30%

MPB1 + KPBO

MPBO + KPBO

Fig. 7: SSAB Oxelösund results showing poorerdesulphurisation for a given alkali output

6. EFFECT OF COATING OF FLUXED PELLETS

The results from EBF campaign 8 are consistent withthe behaviour observed in the full-scale tests. Figure8 shows the relationship of hot metal C and Si for thepellets in campaign 8. The MPB1 pellets clearly showlower carbon contents for equivalent hot metal siliconcontents.

The coating of the MPB1 type pellets appears tosuccessfully alleviate the problem of hot metalquality, with consistently higher carbon content andbetter sulphur distribution and alkali output (Fig. 9).The alkali output was seen to be better for equivalentoptical basicity for coated-MPB1 pellets compared touncoated MPB1 pellets (Fig. 10).

7. DISCUSSION

The particular success of coating of MBP1 pelletscompared to uncoated MPB1 or MPBO appearslinked to two phenomena noted in studies ofmaterials removed from probe samples and from theexcavations:- Clustering- Alkali circulation

4

4,2

4,4

4,6

4,8

1 1,25 1,5 1,75 2[Si]

[C]

MPBO MPB1 MPBO-Quartz MPB1-Olivine

MPB1 MPBO

MPB1-Olivine MPB1-Quartz

Fig. 8: Results of EBF Campaign 8 showing highercarbon content versus silicon for coated MPB1 pelletscompared to both MPB1 and MPBO pellets.

0

20

40

60

80

0,1 0,3 0,5 0,7K2O wt%

(S)/[S]

MPBO MPB1 MPB1-Quartz MPB1-Olivine

MPB1

MPBO

MPB1-Quartz

MPB1-Olivine

Fig. 9. Results of EBF Campaign 8 showing highercarbon content versus silicon for coated MPB1 pelletscompared to both MPB1 and MPBO pellets.

-0,7

-0,6

-0,5

-0,4

-0,3

-0,2

-0,1

0,65 0,66 0,67 0,68 0,69

Optical Basicity

Lo

g (

Sla

g K

2O,%

)

MPB1 MPB1-Quartz MPB1-Olivine

MPB1

MPB1-Quartz

MPB1-Olivine

Fig. 10. Results of EBF Campaign 8 showing higheralkali output for a given optical basicity of slag forcoated MPB1 pellets compared to uncoated MPB1pellets

Clustering of pellets, combined with high meltdowntemperatures could delay carburisation of the ironthat is essential for the lowering of oxygen potential inthe iron and slag.

The clustering of pellets in the blast furnace processhas not received attention. Due to the stability of thedescent of MPB1 pellets, scaffolding and clusteringdo not appear directly related. The reduction andmeltdown conditions of the blast furnace are verycomplex with large amounts of circulating potassium,sulphur compounds, zinc, interaction with otherburden components, as well as temperatures beyondthe melting point of iron. The minerals applied to thepellet surface are likely not the same materials on thesurface at the start of clustering. The interactionsbetween reducing gas, reduction and meltingbehaviour at the pellet surface, effect of coatingminerals on sulphur and alkali distribution in thefurnace and other factors must be considered.

Some alkali behaviours have been studied in theEBF. Generally, alkalis are stable in silicates and butare unstable or unreactive with basic materials.However, the form of the material must be

considered. For example, coarse quartzite does notappear highly reactive, with only a surface reactiontaking place, as shown in the example in Figure 11.In the case of olivine pellets or acid lump ore,potassium appears to be reacting to form K2O-SiO2-FeO slags. For alkali control, silicates (olivine orquartzite) coatings are likely to be effective asindicated in the campaign 8 results in Figure 10. Thecoating particle sizes are less than 100 micron,thereby giving a very high surface area for reaction.

*1*2*3

Relic silica

86% SiO2, 11% K2O, 2% Na2O

33% CaO, 33% SiO2, 19% Al2O3, 5% MgO, 2% K2O, 2% Na2O, 2% S, 1% FeO, 1% MnO.

1 mm

Fig. 11: Reaction of alkali with coarse quartzite in aprobe sample from the start of the cohesive zone ofthe EBF. On the right is quartzite additive, on the leftis a piece of basic sinter.

These preliminary tests of coating of MPB1 pelletswere followed by testing of coated MPBO pellets, withsuccessful results.5

8. CONCLUSIONS

From both pilot -scale and full-scale testing of MPB1pellets, the following conclusions can be made:

1) MPB1 fluxed pellets gives similar production, fuelrate and general blast furnace operation asMPBO pellets.

2) Replacing MPBO pellets with MPB1 pelletsresults in different slag formation behaviour whichaffected the desulphurisation and alkali behaviourin the blast furnaces.

3) Replacing MPBO pellets with MPB1 pelletsresulted in a lower carbon content hot metal for agiven hot metal silicon level.

The behaviours 2-3 were impossible to predict inlaboratory scale, but were detected in experimentalblast furnace trials of 2-3 days for each test material.

From the results of the coating of MPB1 pellets andtesting in the EBF the following can be concluded:

1) Coating MPB1 pellets with 3.6 kg olivine orquartzite improved the desulphurisation andcarburisation of the hot metal, and appeared toimprove the furnace stability and hot metalquality.

2) Coated-MPB1 pellets may be a suitablereplacement for MPBO pellets.

3) The behaviours of the EBF and the full-scalefurnaces were very similar. The EBF is providingreliable evaluations of pellet quality for full-scalefurnaces

9. FUTURE WORK

Full-scale testing of coated-MPBO pellets isunderway at the time of writing.

ACKNOWLEDGEMENTS

We wish to thank SSAB Oxelösund and FundiaKoverhar for their encouragement and for permissionto publish this work.

REFERENCES

1. Sterneland, J.; Hallin, M.: “The Use of anExperimental Blast Furnace for Raw MaterialEvaluation and Process Simulation“, 6th Japan-Nordic Countries Joint Symposium, Nagoya,Japan, November 2000.

2. Dahlstedt, A.; Hallin, M.; Tottie, M.: “LKAB'sExperimental Blast Furnace for Evaluation of IronOre Products“, Proceedings of Scanmet 1, Luleå,Sweden, 1999.

3. Hooey, L.; Sterneland, J.; Hallin, M.: “Evaluationof Operational Data from the LKAB ExperimentalBlast Furnace”, 60th Ironmaking ConferenceProceedings, March 2001.

4. George, D.W.R.; Peart, J.A.: “The Influence ofAlkalis on Blast Furnace Performance“, Alkalis inBlast Furnaces, McMaster Symposium on Ironand Steelmaking, Hamilton, 1973.

5. Sterneland, J.; and Jönsson, P.G.: “The Use ofCoated Pellets in Optimising the Blast FurnaceOperation“,ISIJ, 43 (2003), Nr. 1, p.26-35.