A review of past and present developments of the horizontal single belt casting (HSBC) process

ABSTRACT

Horizontal single belt casting (HSBC) has proven to be a viable future alternative to traditional casting processes (e.g. direct chill and conventional continuous casting for aluminum and steel, respectively). The present paper summarizes HSBC developments in Canada since the 1980s. Theoretical and experimental work is summarized to provide the necessary processing parameters needed to cast a wide range of alloys at pilot and industrial scales. Compared to conventional casting technologies, it is anticipated that HSBC will be a far more versatile, economical, and environmentally friendly method that will also reduce carbon dioxide emissions from metallurgical industries. The effects of air gap dimensions and belt speeds on the stability of the “back meniscus” were also studied for a double-impingement metal feeding system to cast AA2024 aluminum alloy. Using ANSYS Fluent 19.1 computational fluid dynamics software, various combinations of these process parameters were tested to obtain optimum results for promoting back-meniscus stability and assess the effects of these parameters on other phenomena (e.g. air entrainment).

RÉSUMÉ

La coulée horizontale à bande unique (HSBC, de l’anglais horizontal single belt casting) s'est avérée être une alternative future viable aux procédés de coulée traditionnels (par exemple, la coulée à refroidissement direct et la coulée continue conventionnelle pour l'aluminium et l'acier, respectivement). Le présent document résume les développements de la coulée à bande unique horizontale au Canada depuis les années 1980. Les travaux théoriques et expérimentaux sont résumés afin de fournir les paramètres de traitement nécessaires pour couler une large gamme d'alliages à l'échelle pilote et industrielle. Par rapport aux technologies de coulée conventionnelles, le HSBC devrait s'avérer être une méthode beaucoup plus polyvalente, économique et respectueuse de l'environnement, qui permettra également de réduire les émissions de dioxyde de carbone des industries métallurgiques. Les effets des dimensions de l'entrefer et des vitesses de bande sur la stabilité du “ménisque arrière” ont également été étudiés pour un système d'alimentation en métal à double impaction destiné à couler l'alliage d'aluminium AA2024. À l'aide du logiciel de dynamique des fluides ANSYS Fluent 19.1, diverses combinaisons de ces paramètres de processus ont été testées afin d'obtenir des résultats optimaux pour promouvoir la stabilité du ménisque arrière et évaluer les effets de ces paramètres sur d'autres phénomènes (par exemple, l'entraînement de l'air).

KEYWORDS:

• Casting technology, Computational Fluid Dynamics (CFD) modeling, Horizontal Single Belt Casting (HSBC), thin strips/sheets

MOTS-CLÉS:

• bandes/feuilles minces, coulée horizontale à bande unique (HSBC), dynamique des fluides computationnelle (CFD)

INTRODUCTION

The near net shape casting process, horizontal single belt casting (HSBC), was independently conceived in 1988 by R. Guthrie and J. Herbertson at the McGill Metals Processing Centre (MMPC) in Canada and by a research team led by K. Schwerdtfeger at the University of Clausthal in Germany. Since then, the process has been under development as an alternative to conventional processes for producing thin strips of a wide range of alloys. The second emerging near net shape casting technology for steel is twin roll casting or Bessemer casting (Ge, Isac, & Guthrie, 2012). This process has been practiced in the nonferrous industry at relatively low rolling speeds. Attempts by steel companies to apply twin roll casting to ferrous sheet production were unsuccessful until 2008, when Nucor Corp. produced near net shape casting steel sheet products (Ge et al., 2012). Unfortunately, Nucor recently abandoned those commercial operations.

Our HSBC caster was extensively upgraded to process thin strips of high-strength high-ductility, electrical transformer, transformation-induced plasticity and twinning-induced plasticity steels as well as Cu-Sn-Ni and various grades of aluminum alloys, such as AA6111, AA5182, and AA2024. Following the removal of high-temperature equipment from McGill University in 2012, important research was performed at METSIM International’s high-temperature melting, casting, and solidification laboratory in Canada. Crystalline metals and thin strips of aluminum, steel, and copper alloys as well as bulk amorphous sheet products were successfully cast at pilot scale. This led to US and Canadian patents on additional aspects of HSBC. Research at the MMPC and ongoing research and development at METSIM focused on HSBC processing ferrous and nonferrous, crystalline and amorphous 2–10 mm thick strips.

Significant contributions to the development of HSBC technology at the pilot-scale were made by K-H Spitzer at Clausthal University prior to commercialization by Salzgitter AG in 2012 at their steel plant in Peine, Germany, under the name Belt Cast Technology. The research in Germany has focused on processing 10–15 mm strips of various grades of steel. To eliminate hydraulic jumps, the latter approach required electromagnetic damping and gas jets to slow the liquid steel flowing onto the belt.

More recently, HSBC has been used commercially to produce thin strips of copper alloys by Materion Corp., USA. The commercial successes by Nucor, Salzgitter, and Materion illustrate that HSBC and twin roll casting are viable candidates to replace conventional casting methods. Furthermore, HSBC is more cost effective and productive than twin roll casting and is the focus of this paper.

The main advances in near net shape casting until 2014 are summarized in reviews by Ge, Isac, and Guthrie (2013) and Guthrie and Isac (2014). The present publication reviews research since 2014 in Canada on the processing parameters required to cast lightweight aluminum alloys and bulk amorphous metals, namely the type of liquid metal feeding system, belt speed, surface tension, contact angle interactions, and belt surface roughness, all of which can be applied to full-scale industrial processes. Experimental observations and mathematical modeling are discussed to provide the optimal parameters to produce high-quality, thin strips via HSBC.

Experimental work aided by digital twin simulations

At the MMPC, all experimental work at the laboratory scale was validated by experimental work at the pilot scale (Figure 1) for a wide range of metal alloy systems. These pilot studies were accompanied by ANSYS Fluent computational fluid dynamics (CFD) modeling at METSIM International on a dedicated 288 core high-performance computer cluster to enable high-speed computations.

Figure 1. (a) Schematic of pilot-scale horizontal single belt casting and (b) casting an AA2024 alloy strip at METSIM International, Montreal, Canada

The Mark III HSBC simulator (Figure 2) comprises a moving copper rectangular (800 × 110 × 12.7 mm) substrate (Guthrie & Isac, 2014). Large square copper inserts located along its length allow for instantaneous heat flux evaluations for a variety of surface roughness modifications using customized thermocouples with ~10 ms response times. The simulator was extensively used for early research on the effect of substrate topography, substrate texture, and casting atmosphere on cast strip quality. These aspects are fundamental to controlling cooling rates and alloy solidification since gas (e.g., air, helium) entrainment or gas pockets represent an additional layer of heat resistance and thus modify cooling rates. In general, a controlled cast substrate texture/pattern enhanced the interfacial heat flux compared to a sandblasted substrate surface (Guthrie & Isac, 2014; Li, Calzado, Isac, & Guthrie, 2009). The substrate material also significantly affects heat fluxes. For example, heat extraction with a copper substrate is approximately double that with a steel substrate (Li et al., 2006b).

Figure 2. Horizontal single belt casting simulator set-up used by Ge, Isac, and Guthrie (2015b), Guthrie, Isac, Li, and Calzado (2016), and Li, Shabestari, Isac, and Guthrie (2006b) (reprinted with permission from Ge et al., 2015b)

Using the HSBC simulator, a wide range of alloys have been successfully cast to investigate their solidification behavior while forming thin strip products. The MMPC research group also studied the possibility of using HSBC to cast Fe-based bulk metallic glass/amorphous alloys (Guthrie et al., 2016; Li et al., 2006a) to take advantage of the sustained high cooling rate capabilities and horizontal output of bulk metallic glass sheets. Early research focused on casting the following Fe-based bulk metallic glasses (alloys are cited in at% unless otherwise noted): Fe-15B-7Co-7Mo-2Y-8Zr, Fe-6B-15C-15Cr-14Mo-2Y, and Fe-20B-2Zr-2Nb-1.5Cr-4.5 V-2Y. It was concluded that a helium atmosphere is required to enhance the thermal conductance of the interfacial gas layer and obtain the high heat fluxes (8.17 MW/m²) needed to successfully cast this type of bulk metallic glass. In addition, the maximum thickness to promote a completely amorphous matrix was 0.7 mm (Li et al., 2006a).

For casting Ca-Mg-Al–based bulk metallic glasses, Li et al. (2006a) used the HSBC simulator with a perfectly flat copper substrate (Figure 2) to produce 70-mm wide strips of bulk metallic glass. Customized thermocouples were manufactured and included in the HSBC simulator’s cooling substrate to measure instantaneous heat fluxes during solidification of these “glassy” alloy systems. Helium was again injected to displace the air and offer a higher gaseous thermal conductivity, which was expected to enhance heat transfer and obtain thicker amorphous strips. However, helium only increased the maximum heat flux by ~40–70% because some air remained entrapped within the gas mixture within the peaks and valleys onto which the liquid metal was freezing.

To detect recrystallization, X-ray diffraction analysis was performed on all cast samples. For a ternary alloy (60Ca-20 Mg-20Al), an amorphous matrix only formed in the middle region of the 0.8-mm thick strip: no amorphous matrix was observed for the 1.2-mm thick strip. The high cooling rate (800 K/s) required to produce a completely amorphous matrix for the entire strip could not be reached with the HSBC simulator. It was therefore concluded that casting this ternary alloy was not viable via HSBC. However, with a quaternary alloy (60Ca-15 Mg-10Al-15Zn), strips of 0.9 and 1.1 mm thickness were successfully cast, and they exhibited completely amorphous matrices. For a 1.8-mm thick strip, some recrystallization was detected at the edges of the cast strip. It was concluded that the thickness threshold for quaternary alloys was ~1.1 mm. For a quinary alloy (55Ca-15 Mg-10Al-15Zn-5Cu), the maximum thickness to ensure a completely amorphous matrix was 2.0–3.0 mm. In addition, chloride impurities introduced during alloy preparation had to be carefully controlled to avoid the formation of chloride crystals at the grain edges, which compromised strip quality.

The HSBC simulator has also been used to cast the commercial aluminum alloy, AA6111 (Li et al., 2009). Belt surface textures/patterns and casting speeds were explored to optimize processing parameters to obtain a good surface quality for various cooling rates (Figure 3). It was concluded that substrate patterns (d) and (e) yielded the best strip surface quality and an optimal interfacial heat flux by preventing the formation of random air pockets (Figure 4).

Figure 3. Examples of potential copper substrate textures/patterns tested in the Mark III horizontal single belt casting simulator (reprinted with permission from Li et al., 2009)

Figure 4. AA6111 bottom strip surfaces obtained when cast with patterns (d) and (e) from Figure 3 (reprinted with permission from Li et al., 2009)

As noted above, interfacial heat flux—modified by helium gas entrainment between the melt and the surface of the moving belt—is essential to enhance heat transfer during HSBC. Mathematical modeling combined with HSBC simulator experiments proved to be very useful in understanding the evolution of the gas layer with time (Guthrie, Isac, & Li, 2010a, 2010b). In most cases, the heat flux peaked during the initial milliseconds after contact between the molten metal and the cooling surface. It rapidly decreased thereafter, depending on the growth of the gas layer, which occurred in a nonlinear manner from 0 to 6 µm (Guthrie et al., 2010b).

Understanding the intermediate gas layer has been integral in the mathematical modeling of HSBC. In fact, the thermal heat resistance of a 6–10 µm gas layer was used for later CFD models and produced a good fit with experimental results. Modifying the cooling surface texture and type of gas can allow for a wide range of heat extraction rates, which can be customized for various alloys.

Subsequent research at METSIM focused on a novel 5000 series aluminum alloy, Al-Mg-Sc-Zr (Ge et al., 2015b), which offers good corrosion resistance, high ductility, and good weldability. The versatile HSBC simulator was used for experiments, and a mathematical model was developed to validate the simulator results (model details are discussed in the next section). It was concluded that accurate dimensional predictions can be made based on material properties and processing parameters. Moreover, owing to the nonwetting solidification of aluminum alloys on the copper substrate, rounded edges along the length of the strip make it possible to cast without using side dams. The gas entrained between the moving surface and the molten metal contributed significantly to the thermal resistance of the system and to the solidification process. In addition, uniform heat transfer was observed. Initial heat fluxes (~10–20 MW/m²) were higher than typical values for conventional casting (~1–3 MW/m²). These high heat transfer rates and associated high cooling rates improved the microstructure by nucleating smaller grains (Figure 5). They potentially offer many advantages over as-cast alloy strips in terms of microstructure and mechanical properties, provided thickness variances can match those obtained with direct chill cast strip material (Ge, Celikin, Isac, & Guthrie, 2015a).

Figure 5. Measured (±10%) and predicted transient initial interfacial heat fluxes from the melt to the substrate (along the centerline and 35 mm displaced sideways from the centerline); substrate speed during casting was 0.8 m/s (reprinted with permission from Ge et al., 2015b)

The mechanical properties of a Sc- and Zr-containing aluminum alloy obtained from the HSBC simulator were comprehensively evaluated relative to an ingot-cast product after heat treatment (Ge et al., 2015a, 2015b). Sc and Zr promoted precipitation hardening after heat treatment (Ge et al., 2015a). The Sc and Zr contents were slightly lower for the strip (0.39 and 0.05 wt%, respectively) than the ingot product (0.60 and 0.10 wt%, respectively); the content was similar for the main alloying element, Mg. The high cooling rates of HSBC promoted finer grain sizes (~17 µm) than an equivalent ingot-cast product (~50 µm). The strip product required a shorter heat treatment time (0.75 vs 24 h) to obtain equivalent mechanical properties. Since the rapid solidification promoted a higher Sc saturation in the matrix, less bulk Sc content was necessary to achieve the same mechanical properties. This also promoted higher resistance to over-aging. The HSBC sample had fewer polygonal precipitates, which are detrimental to material properties (Figure 6).

Figure 6. Optical micrographs of an (a) ingot-cast and (b) horizontal single belt casting sample (reprinted with permission from Ge et al., 2015a)

Various metal delivery systems have been studied with the HSBC simulator at the pilot scale. Among the first metal delivery systems researched for A356 aluminum alloy was the single (vertical) impingement system (Figure 7a). The HSBC simulator was rebuilt and operated at MMPC using a 150-KVA transformer connected to a 272-kg (600-lb) steel capacity induction furnace. The prepared melt was displaced upward into the low-head metal delivery system using the downward motion of a refractory piston. The metal was then poured onto a 2.6-m long textured steel or copper belt and intensively water cooled at 20–24 m/min. The metal subsequently solidified on the belt into an as-cast 3–10 mm strip according to the mass flowrate of liquid metal onto the belt.

Figure 7. (a) Single and (b) double impingement horizontal single belt casting feeding systems

The pilot-scale HSBC system was used for extensive experimental research and mathematical modeling of aluminum alloy AA6111 casting using the single or double (inclined) impingement metal feeding systems (Figure 7b; Ge et al., 2013). For the double impingement system, the stream of falling molten metal collided first with the inclined refractory to lose kinetic energy and promote a more stable surface, and then with the moving water-cooled belt. The two collisions produced oscillations in the first impingement zone (Figure 8), which can promote air entrainment and instabilities on the meniscus generated in the back gap zone (Xu, Isac, & Guthrie, 2018). Further research is needed to optimize the distance between the inlet nozzle and the inclined wall, the “free fall distance,” to promote a more stable process and avoid air entrainment (Xu, Isac, & Guthrie, 2019). These flow instabilities were absent when a double impingement metal feeding system was used to cast wider (250 mm) thin strips of AA6111 alloy (Figure 9).

Figure 8. Experimental instabilities created by a double impingement feeding system (reprinted with permission from Xu et al., 2018)

Figure 9. Pilot-scale horizontal single belt casting system used to process 7–10 mm thick, 250-mm wide strips of AA6111 aluminum alloy: (a) view toward the feeding system at the beginning of casting and (b) view toward the pinch roll mini-mill during casting (reprinted with permission from Gonzalez-Morales, Isac, & Guthrie, 2023a)

Successful HSBC pilot-scale experimental work and CFD modeling was carried out by MMPC personnel under the auspices of METSIM using the lightweight AA2024 aluminum alloy (Lee, Isac, & Guthrie, 2018). The final strip (Figure 10a) had similar mean roughness values on both the top and bottom surfaces. A fine, uniform equiaxed microstructure (Figure 10b) was obtained across the thickness of the as-cast thin strip.

Figure 10. (a) Photograph showing horizontal streaks on the edges and (b) optical micrograph at 50× magnification of AA2024 alloy strip obtained via pilot-scale horizontal single belt casting (reprinted with permission from Lee et al., 2018)

Significant research was also performed at the METSIM laboratory to process thin strips of AA5182 aluminum alloy on the HSBC pilot-scale system using the double impingement feeding system. The resulting 5.0-mm thick strips (Figure 11a) had acceptable levels of porosity and a mean grain size of 63.1 µm, approximately half the size of the 123-µm grains of a direct chill cast material. In addition, HSBC resulted in desirable secondary phases within the grains—rather than at grain boundaries—due to the uniformly high cooling rates across the strip (Figure 11b). This translated into improved strength and hardness compared to ingot cast products. An isokinetic velocity (inlet velocity=belt velocity) is preferred to reduce surface disturbances during casting.

Figure 11. (a) Photograph and (b) optical micrograph at 50× magnification of AA5182 alloy strip obtained via pilot-scale horizontal single belt casting (reprinted with permission from Hsin, 2019)

More recent HSBC pilot-scale experimental and CFD work was performed to compare thin strip product quality when casting AA6111 alloy (used for the Ford F150 truck) using single and double impingement feeding systems with no side dams. The resulting strips had similar roughness on the top and bottom surfaces without surface indentations on the latter (Figure 12a). The mean grain size was 85 µm, with intermetallic phases distributed at the grain boundaries (Figure 12b). Some pores (dark spots) were also detected, similar to a direct chill cast product (Niaz, Isac, & Guthrie, 2020b). The double impingement system moderated the impact of the incoming melt stream and decreased the velocity of the stream when impinging with the moving belt, generating fewer surface perturbations and no backflow through the back gap according to previous work (Niaz et al., 2020b).

Figure 12. (a) Photograph of surface and (b) optical micrograph at 50× magnification of AA6111 alloy strip obtained via pilot-scale horizontal single belt casting (reprinted with permission from Niaz et al., 2020b)

Some advanced high-strength ferrous alloys and steels, such as transformation-induced and twinning-induced plasticity steels, have also been successfully cast via HSBC. Conventional casting technologies require excessive hot rolling steps following slab formation: the strong work hardening characteristics of these types of steels preclude a one-step casting-rolling procedure. To date, two transformation-induced and twinning-induced plasticity Fe-Mn alloys have been successfully cast on the HSBC pilot-scale machine and simulator at MMPC/METSIM. The 17Mn-4Al-3Si-0.45C (wt%) (Niaz et al., 2020c) alloy exhibited mean yield and tensile strengths of 654 and 880 MPa, respectively, and a mean surface roughness of approximately 0.2 mm. Figure 13 shows their microstructures, including the annealing and deformation twins. The Fe-21Mn-2.5Al-2.8Si-0.08C (wt%; Niaz, Isac, & Guthrie, 2021) alloy had mean yield and tensile strengths of 610 and 950 MPa, respectively, and a mean surface roughness of 0.13–0.16 mm.

Figure 13. Microstructures of horizontal single belt cast strips of 17Mn–4Al–3Si–0.45C (wt%) alloys (reprinted with permission from Niaz, Isac, & Guthrie, 2020c)

Recent work relates to extensive mathematical modeling of back-meniscus stability during HCBC casting using a single impingement feeding system to process an AA2024 alloy. The CFD work focused on analyzing the effect of belt speed, clearance, and contact angle on backflow and meniscus behavior (Gonzalez-Morales, Isac, & Guthrie, 2022). Back-meniscus stability was shown to be of paramount importance to obtain a good quality strip surface (Figure 14). Optimal parameter combinations were predicted to promote a stable meniscus and finally a stable process, avoiding product defects. Further, a nonwetting contact angle (> 90°) was needed to avoid backflow and to promote a stable meniscus. Therefore, surface modification of the moving belt or of interfacial gas composition may be necessary for contact angle modifications during HSBC of some metals.

Figure 14. Influence of back meniscus behavior on strip surface quality (reprinted with permission from Li, Gill, Isac, & Guthrie, 2011)

MATHEMATICAL MODELING OF THE BACK MENISCUS BEHAVIOR

In the present paper, the optimal processing parameters to cast aluminum alloy AA2024 were tested via CFD modeling. A two-dimensional, transient state, isothermal model of the HSBC system with a double impingement feeding system was generated using ANSYS SpaceClaim software (Figure 15). Two air gap sizes and three belt speeds were used in the numerical simulations for a total of six cases (Table 1). The inlet and outlet boundary conditions were also specified. The initial volume fraction of air was set to 1 (at time equal to zero), and the liquid aluminum/AA2024 alloy volume fraction at the velocity inlet (3 mm) was also set at 1 (Table 2). The moving belt condition was defined as a moving wall with only an X positive velocity component; all other walls were defined with a no-slip condition. The properties of the materials, liquid aluminum/AA2024 alloy, and air were considered constant. The model was defined as being transient/unsteady state using a constant time step interval of 1 × 10⁻⁶ s.

Figure 15. Double impingement geometry generated in ANSYS SpaceClaim software for a back-gap size of 1 mm

Table 1. Modified operating conditions for horizontal single belt casting mathematical modeling

Feeding system type	Double impingement
Back gap (mm)	0.5 and 1.0
Belt speed (m/s)	0.1, 0.5, and 1.0

Table 2. Materials properties and operating conditions

Material property	Liquid AA2024 alloy	Air
Density (kg/m^3)	2,300	1.225
Viscosity (kg/m s)	0.001338	1.7894×10^−5
Surface tension (N/m)	0.914	–
Contact angle with moving belt (^o)	120	–
Inlet nozzle velocity (m/s)	1.0

The mesh for the systems was generated using ANSYS Meshing, with cell refinement in impingements zones for better interphase accuracy (Figure 16). The meshes generated were mostly rectangular, with a mean cell size of 2 × 10⁻⁴ m². The mean number of cells for these geometries was 40,000, with an average orthogonality of 0.99.

Figure 16. Mesh generated in ANSYS meshing for (a) 1-mm back gap and (b) refinement detail in the second impingement zone

The volume-of-fluid multiphase model was used to solve the liquid aluminum/AA2024 alloy-air-walls interactions, and the k-ω shear-stress transport turbulence model was used to describe the fluid flow. The use of these models has been validated (Ge et al., 2015b), and the equations and their details can be found in Gonzales-Morales et al. (2022).

Double impingement, 1-mm air gap

The predicted volume fraction contours for the system modeled with a 1-mm backwall gap and a 0.5-m/s belt speed show that at 0.07 s from the start of casting, a small backflow penetrates the back gap zone (Figure 17). By 0.215 s, it is dragged forward for quasi-steady state operation. Similar results were found for 0.1 and 1 m/s belt speeds. None of the three belt speeds generated significant backflow. This result agrees with the experimental and simulation results of Niaz, Isac, & Guthrie (2020a, 2020b) and Xu et al. (2018), who cast AA2024 alloy using a double impingement feeding system.

Figure 17. Predicted liquid aluminum volume fraction contours and velocity vector maps at 0.07, 0.10, and 0.215 s from the start of casting for a double impingement feeding system with a 1-mm back wall gap at 0.5-m/s belt speed

Air entrainment between the molten metal and the inclined back wall is evident at 0.13 s from the start of casting for the three belt speeds (Figure 18). At 0.18 s, air entrainment reaches the back meniscus, where the volume fraction of liquid aluminum drops to approximately 0.8, making it difficult to accurately track the interface. The effect of the air entrainment is enhanced with a higher belt speed, making the back meniscus more unstable. Air entrainment with a double impingement system was observed by Niaz, Isac, & Guthrie (2020b, 2020a) and Xu et al. (2018), though it was attributed to oscillation in the upper free fall zone, not to oscillations in the zone adjacent to the inclined back wall.

Figure 18. Predicted liquid aluminum volume fraction contours at the meniscus zone at 0.13, 0.16, 0.18, and 0.22 s from the start of casting for a double impingement feeding system with a 1-mm back gap at three belt speeds

As expected, a higher velocity is generally reached within the system at a higher belt speed (Figure 19). However, the vectors adjacent to the inclined wall suddenly change direction chaotically when reaching the back gap due to the air entrainment reaching the gap. Consequently, the flow of liquid metal then also exits the system through the back gap zone. This confirms the instabilities generated by air entrainment, which can produce undesirable final strip qualities. Consequently, a double impingement system with a 1-mm back gap is not recommended to process AA2024 alloy—or any aluminum alloy.

Figure 19. Predicted liquid aluminum velocity contours with vector maps within the back-meniscus zone at 0.22 s from the start of casting for a double impingement feeding system with a 1-mm back gap at three belt speeds

Double impingement, 0.5-mm air gap

Predicted volume fraction contours for a 0.5-mm back wall gap and 0.5 m/s belt speed show that no backflow is generated at 0.04 s (Figure 20). Similarly, after 0.15 s, no air entrainment is evident in the presence of a stable back meniscus. This same behavior is observed for 0.1 and 1.0 m/s belt speeds (data not shown).

Figure 20. Predicted liquid aluminum volume fraction contours and velocity vector maps at 0.04, 0.08, and 0.15 s from the start of casting for a double impingement feeding system with a 0.5-mm back wall gap at 0.5-m/s belt speed

For a 0.5-mm gap, no air entrainment is predicted at any of the three belt speeds (Figure 21). Regardless of belt speed, a stable meniscus with no instabilities caused by air entrainment and no oscillations are promoted during quasi-steady operation at 0.15 s.

Figure 21. Predicted liquid aluminum volume fraction contours at the meniscus zone at 0.04, 0.06, 0.08, and 0.15 s from the start of casting for a double impingement feeding system with a 0.5-mm back wall gap at three belt speeds

The velocity magnitude and direction remain unaffected by belt speed (Figure 22). Compared to the 1-mm back gap (Figure 19), the direction of the vectors adjacent to the inclined wall remain constant. Given the considerable improvements over the 1-mm back gap, a 0.5-mm back gap is recommended for casting AA2024 alloys and most other aluminum alloys, assuming the density, surface tension, and viscosity values remain fairly constant. A 0.5-mm back gap promotes back-meniscus stability and prevents air entrainment. As an aside, it is also possible to insert a thin strip of metal foil or metal gauze to stabilize the back meniscus line, if necessary.

Figure 22. Predicted liquid aluminum velocity contours with vector maps within the back-meniscus zone at 0.22 s from the start of casting for a double impingement feeding system with a 0.5-mm back gap at three belt speeds

The “free fall distance” plays an essential role in preventing oscillations and air entrainment (Xu et al., 2019). A free fall distance of 3 mm enhances the stability of the falling stream with no air entertainment (Figure 23) relative to a free fall distance of 6 mm (Niaz, Isac, & Guthrie, 2020b, 2021; Xu et al., 2018).

Figure 23. Double impingement feeding system free fall distance (top) and corresponding calculated melt volume fraction contours (bottom) for AA6111 aluminum alloy, transformation-induced plasticity (TRIP) steel, and AA2024 aluminum alloy

Overall, mathematical modeling shows that a 1-mm back-wall gap causes air entrainment and lowers the back-meniscus stability, regardless of the belt speed. In contrast, a 0.5-mm gap promotes stability of the meniscus and prevents air entrainment at the three belt speeds studied. These predictions confirm the adequacy of a 0.5-mm gap for experimentally casting an AA2024 alloy. They serve as an example of the current research state of HSBC, showing its potential for future industrial applications.

CONCLUSIONS

The present paper summarizes comprehensive research and development undertaken by the MMPC at METSIM’s high-temperature melting and casting laboratory in directly casting thin strips of various ferrous, nonferrous, crystalline, and amorphous alloys via HSBC. It also describes some of the process intricacies of this simple metal processing method (vs direct chill or continuous slab casting) that can produce high-quality thin sheet material directly from molten metal within a fully continuous process.

Several metal alloys have been cast using the HSBC pilot scale equipment (Table 3). HSBC shows several advantages over direct chill or slab cast products. Important equipment and process parameter improvements, like those mentioned in the present paper, can increase HSBC productivity and product quality of ferrous, nonferrous, and copper-based alloys. These findings demonstrate the potential for HSBC to produce high-quality thin metal strips for a broad range of applications. Given SMS-Salzgitter large-scale successful casting operations, future HSBC systems can be envisaged that will be capable of meeting the growing need to reduce liquid metal processing costs and contaminant emissions within a single continuous operation for a metal-hungry world.

Table 3. Chemical composition (wt%) and dimensions of light metal alloys successfully cast using horizontal single belt casting at pilot scale and/or simulator levels

Alloy	Al	Cu	Mg	Mn	Si	Zn	Sc	Zr	Ca	Thickness (mm)	Width (mm)
Ca-Mg-Al	20	–	20	–	–	–	–	–	60	0.8	70
Ca-Mg-Al-Zn	10	–	15	–	–	15	–	–	60	0.9–1.1	70
Ca-Mg-Al-Zn-Cu	10	5	15	–	–	15	–	–	55	0.5–2.0	70
Al-Mg-Sc-Zr	Balance	–	3.93	–	–	–	0.39	0.5	–	5.0–7.0	70, 100
AA5182	Balance	–	4.0–5.0	0.2–0.5	<0.2	–	–	–	–	5.0	200, 250
AA6111	Balance	0.790	0.879	0.287	0.903	0.0007	–	–	–	6.0	250
AA2024	Balance	3.8–4.9	1.2–1.8	0.3–0.9	<0.5	–	–	–	–	8.3	183

REVIEW STATEMENT

Paper reviewed and approved for publication by the Metallurgy and Materials Society of the Canadian Institute of Mining, Metallurgy and Petroleum.

ETHICS APPROVAL AND CONSENT TO PARTICIPATE

There are no ethical issues associated with this manuscript.

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the authors.

The present work is an expansion of two papers (Gonzalez-Morales, Isac, & Guthrie, 2023b, 2023c) published in the Proceedings of the 61st Conference of Metallurgists (COM 2022).

Additional information

Funding

The authors acknowledge financial support from the Natural Sciences and Engineering Research Council of Canada (NSERC) and the Aluminum Research and Development Centre of Quebec (CQRDA), as well as member companies of the McGill Metals Processing Centre. The authors also acknowledge the support in software licensing received from ANSYS Inc. to facilitate this research.

Notes on contributors

D. R. Gonzalez-Morales

D. R. Gonzalez-Morales holds an MSc from McGill University, where he is currently carrying out PhD studies. He is mainly interested mathematical modeling of metallurgical processes, having experience studying complex fluid flow and heat transfer mechanisms in secondary steelmaking and casting processes using ANSYS Fluent. He has published several papers in conference proceedings and refereed journals.

M. M. Isac

M. M. Isac is Associate Director of the McGill Metals Processing Centre (MMPC) with an MEng and a PhD in metallurgy. She enjoyed a 20-year career as a Professor of Physical Metallurgy at the University of Bucharest and a one-year sabbatical specializing in High Resolution Electron Microscopy at Delft University, followed by 28 years of research on process metallurgy at McGill. She has more than 300 publications.

R. I. L. Guthrie

R. I. L. Guthrie is Director of the MMPC. An experienced researcher, he co-invented the LiMCA system for in-situ monitoring of inclusions in liquid metals and the horizontal single belt casting system for casting metals, among many other materials, by working closely with industry. He is the author of the popular textbook, Engineering in Process Metallurgy and co-author of the book, Physical Properties of Metallic Liquids.