RECYCLING OF FLUE DUST INTO THE BLAST FURNACE

Transcription

1 RECYCLING OF FLUE DUST INTO THE BLAST FURNACE Peter Sikström * Lena Sundqvist Ökvist ** * MEFOS Box 812 SE Luleå Sweden psi@mefos.se ** SSAB Tunnplåt AB SE Luleå Sweden lena.sundqvist@ssab.com ABSTRACT Blast furnace flue dust contains valuable amounts of carbon and iron. The contents of harmful trap elements are low enough for recycling into the blast furnace. Since 1993, SSAB Tunnplåt in Luleå has recycled the flue dust by charging it into the blast furnace in the form of a coldbonded dust briquette. In the summer of 2000, the recovery of dry dust increased considerably, when the blast furnace and the gas cleaning system were rebuilt. This in turn influenced the briquette quality and briquette productivity negatively. For this reason, recycling of part of the flue dust by tuyere injection was taken into consideration. The possibility to inject the flue dust was evaluated by characterization of the flue dust and studies on its effect on the raceway conditions. Two successful pilot-scale tests (2 days and 5 days, respectively) were carried out in the LKAB experimental blast furnace. The amount of reducing agents could be decreased and the BF operation was stable. The test results implied that tuyere injection is a suitable method for recycling of flue dust. Introduction In August 2000, SSAB Tunnplåt in Luleå started up a new blast furnace that replaces the two smaller ones. The production capacity of BF No.3 is higher than the combined production capacity of both the former BF No.1 and No.2. The gas cleaning system has also been rebuilt and a cyclone and wet scrubber have been installed. The gas cleaning was previously accomplished by a dust catcher and a wet scrubber. The new gas cleaning facilities have resulted in an increased ratio of recovered dry flue dust. In the old gas cleaning system, approximately 60% of the dust was recovered as dry dust compared to 80% in the new gas cleaning system. All dry flue dust is recycled into the blast furnace as one component in a cold-bonded dust briquette. Dry flue dust has a negative effect on briquette quality in terms of cold strength, and the cost of briquette production has increased. The results from studies made in a laboratory rig and in a charcoal furnace by Gudenau et al. concerning the injection of iron-oxide containing dust has established that hematite can be reduced to metallic iron within the short residence time in the raceway. The experiments also indicated combustion of low volatile coal was promoted by the addition of a certain amount

2 of dust to the mixture [3][4]. The reduction behaviour is dependent on the particle size and chemical composition of the dust [5]. In addition to coke fines, the flue dust also contains fines generated from BOF slag, limestone and pellets, and as a result it contains basic oxides such as CaO and MgO as well as a considerably high amount of Fe 2 O 3. In several studies on injection of fluxes and iron ore, the effect on hot metal composition has been considered. Tuyere-injection of flue dust into the blast furnace might have similar effects as those found in these studies. Injections of fines of fluxes and iron oxide in some blast furnaces to decrease the Si content of hot metal have been tested [6-12]. In a study of simultaneous injection of coal and dolomite, a significant decrease in the Si content of hot metal was seen [7]. The effect of adding material containing MgO from the top or by tuyere injection showed that the silicon content was decreased in both cases. However, the decrease was greater when the flux was injected instead of top-charged. The effect of top charging was more significant if the flux was charged into the coke layers than when it was mixed with ore [6]. A test of tuyere injection of BOF slag into the LKAB Experimental Blast Furnace (hereafter called EBF) also resulted in greatly decreased Si content of hot metal [12]. The reason for injecting fluxes containing CaO and MgO is that the activity of SiO 2 in the tuyere slag is thereby decreased and followed by a decreased generation of SiO gas. The injection of only iron ore powder did not have any notable effect on the Si content of the hot metal, while a mixture of flux and iron ore did decrease the Si content [9]. In another study on injection of iron ore fines, it was found that the Si content was decreased. If sinter fines containing CaO were used the effect was higher [8]. When a mixture of iron ore and coal fines or a water-slurry containing iron ore was injected, decreased silicon content of the hot metal could be noticed [10]. The injection of iron ore dust containing some CaO reclaimed from the storage bins of the blast furnaces had a significant effect in decreasing the Si content of the hot metal. The report also states that it is important to depress the reduction of iron oxide. The different results achieved in the studies on the injection of iron ore fines have been partly explained by the particle size of iron ore [11]. In the case of a very fine grain size of iron ore, reduction in the raceway will be efficient and the effect on the oxygen potential limited. The original oxidation degree of the slag in the bosh and the hearth is also important [9]. Injecting flue dust instead of mixing it in the briquette blend offers several advantages. The strength of the cold-bonded briquette can be improved and as a result the amount of screenedoff fines of briquettes is decreased. It might also be possible to recycle other raw materials in the briquette. An improved gain of the C in the flue dust and improved coal combustion efficiency may result if the flue dust is tuyere-injected. The purpose of this study is to evaluate the possibility of recycling the flue dust into the blast furnace by tuyere injection instead of as one component of the dust briquette. The properties of the flue dust and its effect on coal combustion efficiency have been studied in the laboratory. Two pilot-scale tests a 2- day pre-test and a 5-day test - have been carried out in the LKAB EBF. Two full-scale tests, 5 hours and 48 hours, respectively, have been performed at BF No.3 at SSAB Tunnplåt in Luleå. Recycling of Flue Dust to the BF by a Cold Bonded Briquette kg/thm of a cold-bonded briquette have been charged to the blast furnaces of SSAB Tunnplåt in Luleå since Today, the blend of by-products used for briquette production contains blast furnace flue dust, filter dust from environmental filters, briquette fines and a

3 scrap mixture consisting of coarse particles of BOF sludge and fines of steel and desulphurisation scrap. The ratio of flue dust used has varied between 24-36% of the blend and the results from several studies on improving the briquette strength have shown that the flue dust has a negative effect on the cold strength of the briquette. The reason for the decrease in cold strength when the amount of flue dust increases has been attributed to the content of the coke particles, its pores absorbing water during curing and its surface properties, and the particle size distribution. On the other hand, the C content in the flue dust replaces some of the reducing agents used in the blast furnace. All flue dust generally recycled and the amount of dry flue dust increased from 20 kt in 1999 to 30 kt in 2001, when the reclaim of dry flue dust was increased and the amount of BF sludge decreased. To minimise the percentage of flue dust in the briquetting blend and maintain the cold strength to some extent, production in the briquetting plant has been maximised and the addition of binder increased. The volume of the curing chamber limits the possibility to further increase the production of briquettes. This means that the curing time is decreased when the briquette production increases, which has a negative impact on the cold strength of the briquette. In a study done at SSAB in Luleå, based on a statistical test plan, the replacement of flue dust by screened-off fines from the blast furnace (BOF slag, lime stone and manganese slag), desulphurisation scrap, mill scale sludge and coke breeze was tested. All the recipes tested resulted in increased cold strength compared to when flue dust is added. Additionally, the test results indicate that the amount of binder added can be decreased. Characterisation of the flue dust The flue dust contains fine particles of the blast furnace burden materials. The main components are pellet fines and coke fines, and therefore the flue dust contains approximately 20-30% Fe and 40-50% C. In Table 4, a typical chemical analysis of the flue dust is shown. % Pulverized coal Flue dust 1st test Flue dust 2nd test 0 In an earlier injection trial the flue dust was ground in a rod mill to k80 < 80 µm. In that case it was very sticky and clogged in the pipes. The reason is that the particles that are easy to grind become very fine, before the hard particles have decreased to 80 µm, Figure 1. Laboratory tests have shown that the ground flue dust does not show any water absorption at all, because the porous coke-breeze is ground to a homogenous material. For this test, unground flue dust was chosen, which means that it was taken directly from the cyclone at BF No. 3 at SSAB in Luleå. When the dust is taken from the cyclone, it has gone through a windscreen when it follows the blast furnace off-gas out of the blast Figure 1. Particle size distribution of injected materials. µm Table 1. Flue dust particle size distribution for actual trial Particle size Wt.% of fraction 1000µm µm µm µm µm µm µm µm 10.9 Accumulated wt.%

4 furnace. This means that the particles have the same velocity of fall, independent of density. For fluidisation of flue dust together with coal powder, this is a big advantage. The particle size distribution of the flue dust for this trial is shown in Table 1. The unground flue dust is porous, and it is primarily the coke-breeze in the flue dust that is porous. Because of that, the unground flue dust shows a high water adsorption, but if the unground flue dust has been stored under dry conditions it is easy to fluidise and to transport pneumatically. Due to the content of coke-breeze and iron ore particles, the flue dust causes wear in the pipes. Table 2. Factor and levels used in 75 Blast Temperature BF dust PCR laboratory test. Factor Levels Blast temperature, ºC PCR, kg/thm Amount of BF dust, kg/thm Particle size of BF dust, mesh Particle size of PC, mesh 200 Laboratory tests - The effect of flue dust injection on coal combustion. Nine laboratory tests, each repeated twice, in a fixed bed and performed according to a reduced experimental plan based on the levels of tuyere parameters shown in Table 2 have been carried out in cooperation with University of Science and Technology in Beijing. A preliminary evaluation of the test results shows that the blast temperature, the PC rate as well as the amount of BF dust have a significant effect on combustion efficiency. The effect of an increased blast temperature and BF dust amount on the combustion efficiency is positive, but an increased PCR has a negative effect. The change in the particle size of BF flue dust from 150 to 200 mesh had no significant effect on the coal combustion efficiency. Injection of Flue Dust into the LKAB Experimental Blast Furnace. Technical description of the Experimental Blast Furnace and its injection system A simplified layout of the EBF is shown in Figure 3. It has a working volume of 8.2 m 3 and a hearth diameter of 1.2 m. There are three tuyeres placed at 120- degree intervals. The blast is normally preheated to 1200 C in a new type of pebble heater. Combustion efficiency, % Level of factor Figure 2. Effect of blast temperature, PCR and amount of flue dust on coal combustion efficiency. The EBF is equipped with a bell-type top. A moveable armour is used for the burden distribution control. Two mechanical stock rods monitor the burden descent and control Figure 3. Illustration of the BF and its peripheral equipment

5 the charging of the furnace. The furnace has one tap hole, which is opened with a drill and closed with a mud gun. The hot metal and slag are tapped into a ladle. Probes for temperature measurements, gas analysis and solid sampling over the blast furnace diameter are installed at three different levels. To facilitate dissection and repair, the hearth is detachable and can be separated from the furnace. Operating the Experimental Blast Furnace The blast furnace is operated in campaigns of 4-10 weeks at a productivity ranging from 3.2 to 3.8 t/m³day. The normal tap-to-tap time is 60 minutes and the normal tapping duration 5-15 minutes. Process data are logged continuously and stored in a database. The data are transferred at regular intervals to another database, where reports and trend charts are generated and process calculations are carried out. The EBF is a very sensitive tool for detecting differences in properties of different ferrous burdens. The response time is much shorter for the experimental furnace compared to a commercial furnace. Injection System Figure 4 shows a schematic drawing of the injection system. The EBF is equipped with a lock-hopper coal-injection system. A cylindrical fluidising chamber is fitted below the injection vessel. That chamber fluidises the coal and supplies the pipes with coal for transport to the blast furnace. There is one transport line for each tuyere. For the auxiliary injection, as when flue dust is injected, a separate vessel is connected to the fluidisation chamber, with a volumetric screw feeder. The BF flue dust is mixed together with the coal powder in the fluidisation chamber. Flue Dust Screw Feeder Coal To Blast Furnace Figure 4. Injection system Raw materials The choice of raw materials was based on the raw materials used at SSAB Tunnplåt in Luleå. Olivine pellets - MPBO from LKAB in Malmberget - and coke produced at the coking plant of SSAB Tunnplåt AB in Luleå were used. Pulverised coal from the coal injection plant and flue dust from the gas cleaning equipment of BF No. 3 were used as injectants. Fluxes of the same type as those used at SSAB, except for the quartzite that was added to maintain the slag amount, were charged. Table 3. Raw materials used during the test, kg/thm. Top charging Injection Reference Test Pellets Limestone Quartzite BOF slag Coke Coal Flue dust O 2 enrich. 2.71% 2.75% The coke was crushed and sieved into a fraction of mm, i.e. standard coke for the EBF. The top-charged slag-forming materials were limestone, quartzite and BOF slag. The particles of the slag formers were 9-15 mm in size and the 80% of the particles of the pulverized coal were smaller than 100 µm.

6 Results from the 5 day-test in the Experimental Blast Furnace Injection The injection of flue dust and coal mixture worked well and no problems with the pneumatic transportation occurred. However, as a result of the layout of the injection system, the control of coal injection rate was affected by the addition of flue dust. This also resulted in slightly increased variation in the gas efficiency (EtaCO). The average injection rate during the test was 23 kg flue dust per tonne hot metal. Consumption of reducing agents As can be seen from Table 3, the consumption of reducing agents based on mass balances was 544 kg/thm during the reference period and decreased to 523 kg/thm during the test with injection of flue dust. The injected flue dust had a carbon content of 46.9%. The C content in the injected amount of flue dust corresponds to approximately 14 kg of pulverized coal. A very low amount of this C is present as carbonate. No significant difference in the heat level of the blast furnace was noticed except for a slightly lower Si content of the hot metal during the test period compared to during the reference period, Table 5. Process Both the reference period and the test period were characterised by a stable blast furnace operation in terms of an even burden descent rate and a stable chemical composition of hot metal. However, the variation in burden descent rate decreased significantly when flue dust was injected. The average of EtaCO was 45.9 during the reference period and 46.4 during the test period. As mentioned above, the variation in EtaCO was slightly higher during the test period than during the reference period. The burden resistance index is higher during the test period than during the reference period, but it varied less during the test. The Si content of hot metal was lower during the test period than during the reference period. As shown in Table 5 and in Figure 5, the Si content decreased from 2.18% during the reference to 1.78% during the test. At the Table 4. Chemical composition of the raw materials used. Pellets Lime Quart BOF Flue Coke Coal stone site slag dust Fe 66,8 0,20 0,73 21,4 0,38 0,44 23,4 CaO 0,29 0,99 0,71 41,1 0,020 0,25 7,27 MgO 1,46 1,08 0,46 10,8 0,060 0,11 1,53 SiO 2 2,15 0,51 92,4 8,61 5,73 4,13 5,42 Al2O3 0,41 0,010 3,10 1,39 2,81 1,49 1,70 TiO 2 0,23 0,030 0,090 1,65 0,17 0,060 0,23 V 2 O 5 0,23 0,060 4,71 0,20 Na 2 O 0,034 0,060 0,050 0,020 K 2 O 0,027 0,080 0,88 0,020 0,120 0,150 0,081 S 0,001 0,050 0,020 0,040 0,54 0,73 0,41 MnO 0,049 0,020 0,020 3,79 0,050 1,07 C 90,5 78,8 46,9 Volatile 0,90 39,1 Ash 9,90 6,90 CO 2 43,0 Moisture 0,17 1,0 0,40 0,50 2,30 1,00 0,10 Table 5. Chemical composition of hot metal and slag. Hot Metal Slag wt.% Si Reference Test Temperature 1461ºC 1448ºC %C %Si %S %CaO %SiO %MgO %Al 2 O B Average Si %Si ref %Si test Average C %C ref %C test Reference Test Figure 5. Contents of C and Si in hot metal. wt% C

7 same time, the difference in C content was very small. The C content was 4.49% during the reference period and 4.45% during the test period. Dust generation The generation of flue dust was 5.1 kg/thm during the reference period and 5.0 kg/thm during the test period. The content of solid material in the sludge was lower during the test period than during the reference period. The noticeable changes in the chemical composition of the generated flue dust can be explained by changes in the moisture content of the topcharged burden additives. Full-scale test Two short full-scale tests, 5 and 48 hours long, respectively, have been performed at BF No. 3 at SSAB Tunnplåt in Luleå. To be able to use the ordinary coal injection plant a pre-mix containing 90% raw coal and 10% BF flue dust had to be prepared outdoors, on the ground. The 2 nd injection test was performed two weeks after the preparation of the pre-mix. Since the material had been stored outdoors, it clogged and some problems occurred when it was transferred into the raw-coal silo. The mixture was ground, transferred into the fine-coal silo and finally injected into the blast furnace. The mixture did not cause any significant effects on the particle size distribution or on the performance of the pneumatic transportation as compared to the handling of coal only, but the pressure drop over the filter after the coal mill increased. The chemical composition of the ash of the ground material samples taken indicates when the mixture passes through the mill. The chemical composition indicates that it contains a slightly higher ratio of flue dust than the planned 10%. The total injection rate was approximately 150 kg/thm. As the test was very short and some other disturbances occurred during the test period, is it difficult to draw any conclusions from the process data. Discussion The consumption of reducing agents decreases when flue dust is recycled, because the C in flue dust replaces some of the C in coal and coke. The consumption of reducing agents decreased by 21 kg/thm when a flue dust amount corresponding to a PC rate of 14 kg /thm was injected. The gain of C was very high and the replacement ratio is estimated to be approximately 1.5. The decreased Si content of hot metal can be the main reason for the high gain. The improved coal combustion efficiency achieved when flue dust was added to the coal might also have some effect. The hot metal composition was stable during the injection test. The decreased Si content was probably a result of changed conditions in the raceway. The estimated amount of tuyere slag increases from 29 kg/thm to 32 kg/thm when flue dust is injected. As a result of the content of fines of BOF slag and limestone in the flue dust the basicity of the tuyere slag is increased. The Fe 2 O 3 corresponding to 23.4 wt.% of Fe in the flue dust is supposed to be reduced in the raceway. However, this Fe 2 O 3 will influence the oxygen potential in the raceway. Table 5 shows the estimated chemical composition of the tuyere slag formed based on the assumption of a 100% reduction of the iron oxide present in the raceway. As can be seen, the chemical composition of the tuyere slag is changed and the basicity (B2) increases from 0.02 during the

8 reference period to 0.11 during the test period. The basicities have been calculated according to the following formulas; B2 = (wt.%cao/wt.%sio 2 ) (1) Bell s ratio = (wt.%cao+0,69*wt.%mgo)/(0,93*wt.%sio 2 +0,18*wt.%Al 2 O 3 ) (2) Figure 6 shows the melting behaviour estimated in Chemsage [14-17] by thermodynamic calculations from 1100ºC and intervals of 50ºC up to 1600ºC. The chemical compositions used for the tuyere slag formed during the reference period and test period, respectively, are stated in Table 6. The melting behaviour is improved, when flue dust is injected. For the reference period, a residual solid slag phase is found up to 1450 C; for the test period the corresponding temperature is 1350 C. An estimation of the melting point according to a phase diagram of the system Al 2 O 3 -CaO-MgO-SiO 2 indicates a melting point of 1700 C of the tuyere slag formed during the reference period and of C of the tuyere slag formed during the test period [13]. At the same time, the viscosity is also expected to decrease. The results from the estimations in Chemsage show that the solid phase of both the reference period and the test period contains chemical compounds containing Na 2 O or K 2 O bound to SiO 2 and Al 2 O 3. The alkalis in these compounds are transferred into the gas phase and not into the slag phase under assumed reducing conditions. Constituents, as for example Na, K NaCN and KCN, are found in the gas phase at high temperatures. The recycling of flue dust can be accomplished by mixing it into a cold-bonded briquette, but this study indicates several advantages of recycling it by injection through the tuyeres instead. The negative effects on the cold strength of the briquette can be avoided and the beneficial effects on coal combustion and tuyere slag formation can be reached, at the same time the recycling costs will decrease. No significant effect on the blast furnace process has been noticed during two short full-scale tests (the second of which was approximately 48 hours) made at SSAB in Luleå. However a longer full-scale test should be performed to verify the results from the pilot-scale tests in the LKAB EBF. Conclusions The experience from production and the results from laboratory tests and pilot-scale tests show that recycling of BF flue dust can be done either by mixing the dust into a cold-bonded briquette or by injecting it through the tuyeres. Valuable contents of especially C and Fe are recovered by recycling of the flue dust and the addition of reducing agents to the blast furnace can be decreased. % melt Table 6. Estimated chemical composition of tuyere slag. All amounts in wt.%. Reference Reference Test Temperature, C Test CaO MgO SiO Al 2 O TiO Na 2 O K 2 O MnO B Bells Ratio Figure 6. Estimated melting behaviour of the tuyere slag formed during the reference period and the test period

9 The injection of flue dust improves the coal combustion efficiency as well as the slag formation in the raceway. The changed conditions in the raceway result in a decreased and stabilised Si content in hot metal. As a result, the gain of C will be further increased. Acknowledgements We wish to thank LKAB for giving us the possibility to carry out the test campaign in the EBF. The encouragement from the colleagues, the work done by the personnel participating in the test campaign and the analyses performed by laboratory staff at SSAB and LKAB are also very much appreciated. References [1] T. de Bruin, L. Sundqvist, Briquetting- One Way of Treating By-Products at SSAB Tunnplåt in Luleå., 2 nd International Congress on the Science and Technology on Ironmaking and 57 th Ironmaking Conference Proceedings, Toronto Canada, (1998). [2] L. Sundqvist, K-O. Jonsson, H-O. Lampinen, L-E. Eriksson, Recycling of in-plant fines as cold bonded agglomerates. Committee on Raw Materials- Seminar Proceedings, p 44-60, Brussels Belgium (1999). [3] H.W. Gudenau, H. Denecke, H. Wipperman, Iron Ore and Iron Containing Dust Injection into the Blast Furnace, 2 nd International Congress on the Science and Technology on Ironmaking and 57 th Ironmaking Conference Proceedings, Toronto Canada, (1998). [4] H.W. Gudenau, A. Babich, H. Denecke, S. Yaroshevskii, V. Kochura, Einblasen von Gichtstaub mit Kohlenstaub in den Hochofen. Stahl und Eisen 119, Nr12. (1999) [5] H.W. Gudenau, K. Stoesser, H. Denecke, v. Schemmann, Environmental Aspects of Recycling of Filter Dusts by Direct Injection or Use of Agglomerates in Shaft Furnaces. ISIJ International, Vol. 40, No.3, (2000). [6] T. Sawada, R. Nakajima, S. Kishimoto, H. Hotta and K. Ishii, Flux injection from the tuyere for low silicon operation. Ironmaking conference proceedings 1990, pp [7] C. Yamagata, Y. Kajiwara, S. Suyama, K. Sato, S. Komatsu, Simultaneouns Injection of Pulverized Coal and Dolomite into Blast Furnace Tuyeres. ISIJ International, Vol. 30, 1990, No. 5, pp [8] K. Kushima, M. Naito, K. Shibata, H. Sato, H. Yoshida and M. Ichida, Iron ore injection into blast furnace raceway. Ironmaking conference proceedings 1988, pp [9] K. Takeda, Y. Sawa, S. Taguchi, N. Takashima, T. Matsumoto and H. Obata, Iron ore and flux injection through blast furnace tuyeres and its effect on raceway phenomena. Proceedings of The Sixth International Iron and Steel Congress 1990, Nagoya, ISIJ. [10] G. Brun, R. Nicolle, J.M. Steiler, J.L. Bouttement, J.L. Eymond, J.P. Menaut, Injection de minerai au haut fourneau et amitrise de la teneur en silicium. La revue de Metallurgie CIT, janvier [11] C. Yamagata, Y. Kajiwara, S. Suyama, T. Miyake, Desiliconisation Reaction of Pig Iron with High FeO containing Blast Furnace Slag under Pressurised and Cokecoexisting conditions, ISIJ International, Vol. 30, 1990, No. 9, pp [12] P. Sikström, L. Sundqvist Ökvist, J-O Wikström, Injection of BOF slag through the Blast Furnace Tuyeres-Trials in an Experimental Blast Furnace, 61 st Ironmaking conference proceedings, Nashville USA, (2002).

10 [13] Slag Atlas 2nd Edition, Verein Deutscher Eisenhüttenleute (VDEh), Verlag Stahleisen GmbH, (1995), , , 381. [14] G. Eriksson, K. Hack, Chemsage A computer program for the calculation of complex oxide equilibria, Met. Trans. B, 21B, (1990), pp [15] A. Hauck, O. Knacke. Conventional Chemical Potentials of Elements in Dilute Ferrous solutions. Steelresearch Vol. 56 (1985) No 10. pp [16] PJ. Spencer, K. Hack. The solution of Materials Problems Using the Thermochemical Databank System THERDAS. Swiss Plastics, vol. 13, No. 9 (1991) pp.63-64, 66, [17] H. Gaye, J. Welfringer, Modelling of the thermodynamic Properties of Complex Metallurgical Slags. 2nd International Symposium on Metallurgical Slags and Fluxes. (1984)