Effect of Melting and Casting Conditions on Aluminium Metal Quality

Transcription

1 Effect of Melting and Casting Conditions on Aluminium Metal Quality Derya DISPINAR, John CAMPBELL Department of Metallurgy and Materials, University of Birmingham, UK. Abstract A study in a secondary alloy ingot producing plant was targeted to investigate the metal quality change in the holding furnace during secondary remelting of aluminum alloy LM24 (Al-8Si-3Cu-Fe). The investigation was an attempt to understand the effect of (i) the use of diffusers (porous plugs integrated into the body of the furnace) and (ii) casting techniques involving different degrees of turbulence. Casting density and bifilm index were found to be useful parameters to assess metal quality. It was found that the metal quality was increased significantly, and was maintained high throughout the casting operation when (i) diffusers were used and (ii) turbulence was reduced to a minimum. Keywords: secondary remelting, aluminum, metal quality, bifilm index Introduction The cost of recycling of aluminium compared to the cost of primary aluminium is highly attractive as a result of major energy savings [1, 2]. Recycling also has benefits for the environment and for the conservation of natural resources. Today recycled aluminium accounts for one-third of aluminium consumption world-wide [1]. The ultimate goal of the recycling process is to produce clean aluminum to an accurate chemical specification while minimising metal losses [3-6]. For this purpose, in addition to alloying, fluxing and degassing treatments are carried out on the metal in the liquid state in melting furnaces. The melt is then transferred to a holding furnace. Finally, the melt is cast into convenient-sized ingots (pigs) that serves as melt stock for the shaped castings industry. These ingots, formulated according to recognised national or international specifications, go into the manufacture of aluminium cast components. The specifications do not, however, cover the amount of oxides or gas that the ingots may contain. These constituents are suspected of being of central importance to the quality of the final cast products. 213/1

2 In an earlier study [7], the use of different lances and ceramic diffusers in an induction furnace during fluxing and degassing was investigated. In addition, different gas flow rates were also examined. It was found that when high gas flow rates were used, violent surface turbulence was observed on the surface of the melt that introduced bifilms [8]. When low flow rates were used, then the cleaning process was found to be less efficient. Thus the control of the remelting process is not straightforward. In addition, there were still some concerns about the introduction of bifilms during transfer of the melt from melting furnace to holding furnace, mainly due to the heavy turbulence that occurred as a result of the high fall of the liquid. It is well known that in controlling the concentration of inclusions in aluminium melts, the settling of the melt has been found to be beneficial to metal cleanliness [9]. Therefore, in this study, the aim was targeted to investigate the metal quality change in the holding furnace as a function of time during a pour. In particular, an attempt was made to understand the effect of diffusers and their importance on the metal quality. In addition, the melt was sampled at various points along the production line, from the tapping of the holding furnace until the mould filling station. Experimental The process begins with the melting of a charge of selected scrap in a 25 kg electric furnace (Figure 1) to produce a melt at 75 o C. Following fluxing and degassing operations, molten metal is transferred to the holding furnace. Two ceramic diffusers sited in the bottom the holding furnace purge nitrogen gas through the melt for a minimum of 2 minutes. During the tests, different numbers of diffusers were run;, 1 and 2. The schematic layout of the holding furnace and position of diffusers is given in Figure 2. In the second part of this same study, three changes were made in the casting area. These are summarised in Figure 3. The first change was the lowering of the height of the launder so that the fall of the liquid metal into the ingot mould would be minimised (Figure 3a 1 ). The second change was made to the casting device that was inted to deliver the melt into the ingot mould at the lowest possible point. The internal geometry of this device (unfortunately not able to be revealed here for commercial reasons) was clearly not altogether satisfactory, involving some degree of turbulent fall. Therefore the design was altered slightly and the rate of production of castings was slowed down. These actions eliminated the worst aspects of the turbulent flow in the device such that the melt was filled from the bottom of the ingot mould with much reduced disturbance (Figure 3a 2 ). The third change was made in the tapping procedure of the holding furnace (Figure 3b 3 ). When not controlled carefully, the liquid metal ted to jet out from the tap-hole with the result that the melt was 213/2

3 violently and turbulently propelled along the launder, clearly creating new bifilms and dross. The study was performed with alloy LM24 (Al-8Si-3Cu). The composition of the alloy is given in Table 1. Three ingots were taken from each casting trial: one from the (when operations in the holding furnace are complete, and the furnace is tapped to the production run), one from dle and one from the (when the holding furnace is empty) of the casting process. These ingots were then separately remelted in an induction furnace at 75 o C and reduced pressure test samples were taken from the melt. (The taking of ingot samples was convenient, allowing the reduced pressure test to be carried out later at a remote location. Furthermore, this approach was not found to affect the RPT result to any detectable extent.) RPT samples (5 x 35 x 15 mm dimensions) were cast into sand moulds bonded with 1.2% resin and solidified at mbar. Archimedes Principle was used to determine the density of the samples. The samples were then sectioned and polished for image analysis and bifilm indices [1, 11] of each sample were measured. Results The density of the RPT samples as a function of the number of diffusers is given in Figure 4. When no diffuser was used, the densities of the reduced pressure test samples fell from 262 to 2454 kg/m 3 as the casting process progressed (Figure 4a). Once the diffusers were active, the density stayed high between -255 kg/m 3 and the results were less scattered (Figure 4b, c). This is more clearly illustrated in Figure 4d when averages of the all results are resented in the same graph. The sectioned surface of the RPT samples can be seen in Figure 5a. The bifilm index results of the same samples are given in Figure 6. At the of the casting, bifilm index (the total length of defects seen on the polished surface of the RPT sample) lies between -15 mm. When no diffusers were operated the bifilm index rose toward 25 mm as the of casting was approached whereas when diffusers were operational bifilm index fell to around mm. It is interesting to note that there is no clear difference between the results when using one diffuser or two (Figure 6d). The second study of the action of the diffusers were carried out after the changes in the casting area were made. The experimental conditions are summarised schematically in Figure 3. The results demonstrate a clear increase in the quality of the castings when these changes were made. The density increases in the reduced pressure test samples are illustrated in Figure 7. In these figures, the term non-quiescent indicates the set of results from the original design, and 213/3

4 the term quiescent indicates the new results after the changes have been made as described in the experimental section. When no diffusers were used (Figure 7a), the increase in the density was approximately + kg/m 3 for each section of the casting; from 262 kg/m 3 to 274 kg/m 3 at the, 2538 kg/m 3 to 2673 kg/m 3 at the, 2454 kg/m 3 to 2617 kg/m 3 at the. The average increase in the density when diffusers were applied (Figure 7b, c) was about half of this value, approximately +5 kg/m 3. The change in the bifilm index results from the improved filling conditions are illustrated in Figure 8. The average bifilm length values for nonquiescent conditions (which is the original fast and turbulent casting) were 188 mm without diffusers and mm with diffusers. In quiescent conditions, these rather high results fell to 39 mm without, and 35 mm with diffusers (Figure 8). These significant improvements can also be seen from the sectioned surface of the RPT samples in Figure 5 a and b. Discussion In an earlier investigation of the quality of the melt in the holding furnace during secondary remelting process [7], RPT sample collection was carried out after 3 and 6 minutes during which an increase in the density of RPT samples and a decrease in the hydrogen content were observed [7]. The positive effect of holding time was possibly due to the settling of oxides [9]. However when comparing the RPT results from to the the casting process, it was found that the results were not consistent. The results of the density change of RPT samples collected from the holding furnace when different numbers of diffusers were run are shown in Figure 4. As seen from Figure 4, when diffusers were not used (Figure 4a), the density of the RPT samples decreased from to the of the casting. This appears to suggest an increased concentration of inclusions in the lower levels of the melt. It is important to note that the chemical composition of the melt remained unchanged during this period (Table 1). However, once the diffusers were used (Figure 4 b, c), the density results remained practically unchanged from the to the. As was expected, the inclusion concentrations were significantly lowered and were, perhaps, uniformly distributed throughout the melt and through the casting process. The Bifilm Index of these samples are in good agreement with the density results: it increased (Figure 6a) from to and once the diffusers were active, the bifilm index decreased dramatically (Figure 6 b, c). For zero, one and two diffusers, the average bifilm index values were 188, 11 and 12 mm, respectively. 213/4

5 In a second phase of these studies, turbulence during the transfer of the melt was reduced at two locations (i) at the outlet of the holding furnace where a stopper was introduced to control jetting, thereby transferring the melt more quiescently; and (ii) the fall of the liquid from the launder into the moulds was reduced to a minimum. After these changes the tests were repeated, with impressive results. There was a clear increase in the quality of the castings. The density of the RPT samples was increased dramatically (Figure 7). The change in the bifilm index was also remarkable (Figure 8). The average bifilm length values for the original non-quiescent conditions were 188 mm without diffusers and mm with diffusers. These rather high results fell in the improved quiescent conditions to 39 mm without, and 35 mm with diffusers. The clear separation of the results is noteworthy: in nonquiescent conditions the average bifilm length was always above 7 mm whereas for quiescent conditions all results were below 6 mm. It is interesting to note that when the casting was carried out quiescently (achieved simply by implementing simultaneously all the three changes investigated in this study), the average bifilm index values of quiescent conditions was below 5 mm even when no diffusers were used (Figure 8). Thus the effect of the diffusers was of somewhat less importance than the control of turbulence, confirming the importance of the control of turbulence during the handling and transfer of molten aluminium alloys [8]. One might speculate that given a careful design of plant, theoretical densities and zero bifilm length might be achievable. Conclusions 1. When no diffusers were used in the holding furnace there was a deterioration in the quality of melt from the beginning to the of the casting. 2. Metal quality increased significantly when one diffuser was used; however there was marginal additional advantage at the limits of detectability when using two diffusers. 3. The more quiescently the casting was controlled, the higher the quality of the products. Good control includes (i) careful minimisation of turbulence at tapping; (ii) minimised fall of the liquid; and (iii) filling conditions to reduce turbulence in the mould. 213/5

6 References 1. International Aluminium Institute European Aluminium Association. 3. Tenorio, J. A. S., Carboni, M. C. and Espinosa, D. C. R., Recycling of aluminum - effect of fluoride additions on the salt viscosity and on the alumina dissolution. Journal of Light Metals, 1. 1(3): p Samuel, M., A new technique for recycling aluminium scrap. Journal of Materials Processing Technology, (1): p Khoei, A. R., Masters, I. and Gethin, D. T., Numerical modelling of the rotary furnace in aluminium recycling processes. Journal of Materials Processing Technology, : p Davies, S. B., Masters, I. and Gethin, D. T. Numerical modelling of a rotary aluminium recycling furnace. in Proc. 4th International Symposium on Recycling of Metals and Engineered Materials.. TMS. 7. Dispinar, D. and Campbell, J., Metal quality studies in secondary remelting of aluminium. Journal of Institute of Cast Metals Engineers, (3612): p Campbell, J., Castings. 2nd ed. 3, Oxford: Butterworth- Heinemann Ltd. 9. Martin, J.-P., Dube, G., Fray, D. and Guthrie, R., Settling phenomena in casting furnaces: A fundamental and experimental investigation. Light Metals, 1988: p Dispinar, D. and Campbell, J., Critical Assessment of Reduced Pressure Test: Part I: Porosity Phenomena. International Journal of Cast Metals Research, 4. 17(5): p Dispinar, D. and Campbell, J., Critical Assessment of Reduced Pressure Test: Part II: Quantification. International Journal of Cast Metals Research, 4. 17(5): p Acknowledgements The authors would like to gratefully acknowledge the financial support of Norton Aluminum, and for their assistance in the use of facilities in the foundry. The diffusers used in the holding furnace were provided by Capital Refractories Ltd. Tables Table 1: Chemical analysis of the 2.5 ton melt from to LM24 Cu Mg Si Fe Mn Ni Zn Pb Sn Ti Al Start rem. End rem. 213/6

7 Figures Figure 1: a schematic representation of aluminium ingot production Figure 2: The schematic drawing of the holding furnace and the position of diffusers 213/7

8 (a) Figure 3: Schematic illustration of changes made at the casting trials (a) 1 The launder was lowered to be as close to the casting mould possible, 2 The casting and filling speed were decreased 3 More care was taken to avoid the severe turbulence on tapping (a) difs 1 dif dif (c) (d) Figure 4: Density change of RPT samples (a) no diffusers were run one diffuser were on (c) two diffusers were on (d) average values of density change of RPT samples 213/8

9 (a) Figure 5: Sectioned surface of RPT samples from the trials (a) non-quiescent castings conditions quiescent conditions (a) dif 1 dif 2 difs (c) (d) Figure 6: Bifilm Index comparing different number of diffusers (a) no diffusers were run, one diffuser were on, (c) two diffusers were on (d)averages of Bifilm Index comparing different number of diffusers 213/9

10 Non Quiescent Quiescent Non Queiscent Quiescent (a) Non Quiescent Quiescent (c) Figure 7: The comparison of the average density values of RPT samples for different techniques (a) no diffusers were run, one diffuser were on, (c) two diffusers were on Non Quiescent Quiescent Non Quiescent Queiscent (a) Non Quiescent Quiescent (c) Figure 8: The comparison of the average bifilm index values of RPT samples for different techniques (a) no diffusers were run, one diffuser were on, (c) two diffusers were on 213/1