Effect of In Situ Al Roll Coating on Strip Surface Quality in Traditional Twin-Roll Casting of Aluminum Alloys

Abstract

The twin-roll casting (TRC) process is widely used in the aluminum industry due to its cost efficiency and continuous production capability. However, maintaining consistently high surface quality remains challenging due to complex heat transfer behavior at the roll/strip interface. This study examines the critical influence of roll surface conditions, especially the formation of an Al coating layer, on solidification behavior and resulting strip quality in the TRC of an Al-5Mg alloy. Experimental results demonstrated that casting without an Al coating layer led to surface defects such as hot tears and porosity due to insufficient cooling. In contrast, strips produced with a stable Al coating layer exhibited excellent surface quality with no surface defects. Numerical simulations further indicated that a stable Al coating enhanced the interfacial heat transfer coefficient (up to 30,000 W/m²K), ensuring effective cooling and complete solidification before the strip exited the roll nip. Moreover, simulations validated the feasibility of using steel rolls in industrial applications, provided the coating layer was consistently maintained. This research highlights the significance of roll surface control in improving TRC product quality.

1. Introduction

Twin-roll casting (TRC) has been utilized in the aluminum industry for over 70 years as a strip manufacturing process [1]. In this process, the molten metal directly contacts the casting rolls, leading to rapid solidification, allowing the direct production of thin strips with a thickness ranging from 3 to 10 mm [2]. As a result, compared to conventional slab-based hot rolling processes, TRC offers significant cost reductions and requires lower capital investment, making it more accessible to medium- and small-sized enterprises [1]. Historically, this process has been predominantly applied to the production of low-alloy aluminum strips, such as the 1000 and 3000 series alloys [3,4,5]. Recently, however, research and development efforts have intensified to expand its application to high-strength aluminum alloys for use in the transportation sector [6,7,8,9,10,11].

The traditional TRC process for aluminum strip production, as shown in Figure 1, primarily employs a horizontal or slightly inclined caster [1]. In this setup, two casting rolls with internally circulating cooling water are arranged as depicted in Figure 1, enabling molten metal to be continuously supplied in the horizontal direction. The casting rolls are typically made of steel shells; however, copper-shell rolls are also used to enhance productivity in mass production. To prevent sticking between the strip and the roll surface during the continuous casting process and to ensure stable strip production, a parting agent is continuously sprayed onto the rotating roll surface. The as-cast strips are continuously coiled and subsequently subjected to heat treatment and cold rolling to obtain the final thin sheets. Compared to the conventional slab-based hot rolling process, this approach significantly reduces the number of processing steps. However, it also restricts the ability to control defects and microstructure, making rigorous quality control at the casting stage essential [8,12].

One of the most critical issues in the TRC process is maintaining the surface quality of the strip [8,13]. In conventional direct-chill (DC) casting, thick slabs of approximately 500–600 mm in thickness are produced. As the alloy content increases, an inverse segregation layer typically forms on the slab surface [14]. To ensure surface quality, these slabs undergo a surface scalping process before hot rolling. In contrast, aluminum strips produced by the TRC process are typically less than 10 mm thick, making surface scalping impractical. As a result, defects such as inverse segregation [8] and surface microcracks [15] cannot be mitigated in subsequent processes, necessitating stringent control during the casting stage. This issue is particularly critical when manufacturing high-strength aluminum alloys with a high alloy content and a wide solidification temperature range, as they are more susceptible to surface defects [8,13]. Therefore, ensuring high surface quality in the TRC process requires advanced casting technology with precise process control.

In the TRC process, various factors influence the solidification behavior and defect formation in strips, including roll rotation speed, melt supply temperature, roll load, set-back distance, roll gap, and other process parameters [2]. Numerous previous studies have investigated the effects of these process variables on defect formation [6]. However, in traditional TRC processes, a more fundamental prerequisite must be addressed before optimizing these parameters. In aluminum strip production, a wide strip width is generally required to enhance applicability and productivity [8]. Since strips produced by industrial TRC are not only thin (typically less than 10 mm) but also require widths exceeding 600 mm, the thickness-to-width ratio becomes significantly high. Consequently, ensuring uniform thickness and consistent quality across the entire width is crucial. This necessitates precise solidification control to achieve uniform solidification across the strip width. To achieve this, it is essential to minimize temperature variations across the width of the melt supplied through the ceramic nozzle [16]. A uniform temperature distribution facilitates even growth of the solidifying shell, enabling the production of wide strips with consistent properties. Additionally, in cases where strip width is particularly large, a crown configuration may be applied to the rolls to optimize roll load distribution and maintain uniform thickness across the strip [17].

In the TRC process, in addition to the temperature distribution of the molten metal, the condition of the roll surface is another critical factor. The traditional aluminum TRC process is designed to integrate rapid solidification and hot rolling into a single operation. As a result, high roll loads are applied to the strip during continuous casting. Unlike conventional casting or DC slab casting, where solidification occurs without significant external stress, the TRC process facilitates the formation of an Al coating layer on the roll surface due to the hot rolling effect under high load conditions. It is important to emphasize that this Al coating layer is fundamentally different from graphite-based parting agents that are intentionally sprayed onto the roll surface during TRC [18]. The formation of an Al coating layer on the roll surface modifies the interfacial heat transfer conditions between the molten metal and the roll surface during solidification. Since solidification progresses continuously as the casting rolls rotate, variations in roll surface conditions can significantly affect the subsequent solidification behavior, necessitating precise and careful control. However, despite numerous studies conducted on the development of traditional TRC processes, limited research has specifically addressed roll surface coating conditions, as they have largely remained a matter of practical field know-how. Without stabilizing the fundamental roll surface coating conditions, controlling numerous other process parameters—such as roll speed, melt temperature, and set-back distance, as well as other process parameters—cannot effectively optimize the process. Therefore, a thorough understanding of these coating conditions is essential for producing high-quality aluminum strips.

The objective of this study is to elucidate the influence of naturally formed Al coatings on roll surfaces on the solidification behavior and resulting surface quality of strips in twin-roll casting. By integrating experimental observations with numerical simulations, this study aims to bridge practical industrial observations with a fundamental understanding of interfacial heat transfer behaviors in the TRC process. Although the TRC process has been extensively studied for decades, few studies have addressed the in-process evolution of roll surface conditions, particularly the spontaneous formation of Al coating layers. This study highlights a rarely documented phenomenon that has direct implications for strip quality and process stability.

2. Materials and Methods

2.1. Strip Fabrication Using Twin-Roll Casting Process

An Al−5Mg (Al-5.0Mg-0.03Si-0.13Fe-0.02Ti, in wt.%) alloy ingot was melted in an electric furnace, and degassing was carried out using Ar gas for 10 min prior to casting. A horizontal-type twin-roll caster was used for strip production. The casting rolls had an outer diameter of 300 mm, and the outer shell of the rolls was made of Cu-1Cr with a thickness of 10 mm. To maintain consistent cooling performance, the initial roll surface was polished with #1000 sandpaper. The initial roll gap and set-back distance were set to 3 mm and 30 mm, respectively. Casting was conducted at a roll speed of 3 rpm, and a 100 mm-wide strip was produced to minimize the effect of temperature variations across the strip width. No parting agent was sprayed onto the roll surface during continuous casting. In the TRC process, the strip surface temperature was measured immediately after exiting the caster using a surface-contact K-type thermocouple. Temperature recording was conducted at a frequency of 2 measurements per second, and temperature variations under each condition were approximately ±5 °C. The strip thickness was directly measured at each location using an electronic thickness gauge. The surface microstructure of the as-cast strip was observed using scanning electron microscopy (SEM, JEOL JSM-6610LV, Tokyo, Japan).

2.2. Simulation Model for Twin-Roll Casting Process

Thermal-fluid simulations were conducted to investigate solidification behavior during the TRC process. The simulations were performed using the commercial software ProCAST (ver. 2021). The model geometry and the boundary conditions are given in

Table 1. Model geometry and boundary conditions.

Condition		Value
Initial melt temperature		670 °C
Initial roll temperature		25 °C
Ambient temperature		25 °C
Interfacial heat transfer coefficient, h	hmelt/nozzle	Adiabatic
	hstrip/air	12 W/m2K
	hmelt/casting roll	Variable

. In general, mesh size in finite element (FE) analysis affects computational accuracy, with smaller mesh sizes leading to higher precision [19]. To balance accuracy and computational efficiency, this study implemented a variable mesh size strategy, applying finer meshes in critical solidification regions. For the critical solidification region in the TRC simulation, the smallest computationally feasible mesh size of 0.1 mm was applied to ensure precise modeling of the solidification behavior. Since ProCAST software fundamentally supports only 3D analysis, a pseudo-2D simulation was performed by initially generating a 2D mesh for the overall geometry and subsequently extending it by a single-layer mesh in the depth direction. The calculation was performed until a steady-state condition was reached, where no further temperature variations were observed in the strip and solidification regions. Figure 2 presents the thermo-physical properties of the Al-5Mg alloy used in the simulation. These material properties were obtained from the PanAl2021 database integrated into the ProCAST software. The data were calculated under Scheil cooling conditions to account for the high cooling rates characteristic of the TRC process.

3. Results and Discussion

3.1. Effect of Roll Surface Condition on Solidification Behavior in TRC

Before analyzing the solidification behavior on the roll surface in the TRC process, it is crucial to first understand the initial solidification behavior at the mold surface in conventional casting processes. Figure 3 illustrates the formation of an initial solid skin at the mold surface in typical casting processes [20]. When molten metal contacts the mold surface, heat is rapidly extracted from the relatively hot molten metal to the cooler mold, leading to solidification. Nucleation and subsequent growth lead to the formation of an initial solid skin. A magnified view of the interface between the molten metal and the mold reveals that the two surfaces do not achieve perfect contact; instead, trapped air pockets partially form at the interface. Heat transfer through these air pockets is inherently less efficient than that at the direct contact interface between the molten metal and the mold. As a result, heat transfer occurs through both direct contact at the interface and conduction across the trapped air pockets. The efficiency of this heat transfer is influenced by several factors, including mold surface roughness, molten metal pressure, wettability, and the presence of a lubricant.

In contrast, although the initial solidification behavior in traditional TRC processes follows a similar mechanism, the roll surface conditions differ significantly from those in conventional casting methods. During continuous casting in TRC, high roll loads exerted on the strip induce the spontaneous formation of an Al coating layer on the casting roll surface. Figure 4 presents the surface appearance of the casting rolls used in this study before and after continuous casting. As illustrated in the schematic diagram, a continuous Al coating layer forms on the roll surface where it is in direct contact with the strip. A crucial technical aspect to note is that this coating layer only develops when a sufficient roll load is applied to induce a hot rolling effect.

Figure 5 presents the results of strip production in the TRC process under conditions with and without the formation of an Al coating layer on the roll surface. Both strips were produced under identical casting conditions, including casting speed, with the only difference being the presence or absence of the coating layer on the roll surface. As shown in Figure 5, when the strip was cast without a coating layer on the roll surface, the surface exhibited a dull and non-uniform appearance. In contrast, the presence of a coating layer resulted in a more uniform strip surface, closely resembling that of a conventionally rolled strip.

The thickness distribution across the strip width is shown in Figure 6. The most notable difference between the two conditions is the strip thickness. In the TRC process, the strip thickness is fundamentally determined by the combined thickness of the solidified shells growing on both roll surfaces. Despite being cast under the same process conditions, the strip produced without a coating layer had a thickness of approximately 3 mm, nearly identical to the initially set roll gap. In the TRC process, the strip thickness must exceed the initial roll gap to facilitate roll gap expansion, thereby enabling the application of roll load to the strip. Thus, this result suggests that minimum roll load was applied. Conversely, the strip produced under conditions with an Al coating layer exhibited a greater thickness of approximately 4.8 mm. This result indicates that, under casting conditions with a coating layer on the roll surface, the contact between the solidifying shell and the roll surface was improved, enhancing the cooling rate and leading to the formation of a thicker solidified shell. In conventional solidification processes, the formation of an artificially applied coating layer on the mold surface generally reduces the interfacial heat transfer coefficient (HTC) [21]. This reduction is attributed to the increased interfacial resistance caused by the coating layer itself and the lower thermal conductivity of the coating material [22]. However, in the TRC process, the naturally formed coating layer, composed of the same material as the cast strip, suppresses the formation of air pockets between the roll surface and the molten metal. This improves wettability and enhances overall heat transfer efficiency.

Figure 7 presents the surface microstructures of the strips produced under the two conditions shown in Figure 5. For the strip with a dull and non-uniform surface, dendrite branches were clearly observed, along with a significant presence of hot tears and porosity near the surface, as shown in Figure 7a. This suggests that residual liquid remained on the strip surface as it exited the caster, leading to slow solidification under air cooling conditions. In contrast, the strip with larger thickness and a uniform glossy surface exhibited no casting defects and had a smooth overall appearance (Figure 7b). This indicates that the strip surface was sufficiently cooled to below solidus temperature before exiting the caster, allowing the solid shell to develop fully. Under these conditions, the solid shell was subjected to roll load, promoting the formation of a smooth surface.

3.2. Effect of Interfacial HTC on Solidification Behavior in TRC Simulation

In the previous section, the effect of roll surface conditions on the surface quality of TRC strips was investigated. Notably, despite identical casting speeds and set-back distance conditions, significant variations in strip thickness were observed. This indicates that the cooling efficiency differed between the two roll surface conditions. In this section, a simulation model is employed to examine the influence of the interfacial HTC between the molten metal (or solidifying shell) and the casting roll on the overall solidification behavior in the TRC process. Figure 8 presents the simulation results predicting strip temperatures under different HTC conditions. The simulation results show that as the HTC value increased, the strip temperature decreased. To analyze the two cooling conditions presented in Figure 4, the strip surface temperature was measured immediately after exiting the caster during the TRC process using a contact-type thermocouple. The experimentally measured strip temperatures are represented by the orange-colored bars in Figure 8. Based on the experimental results, the estimated average HTC values were approximately 8000 W/m²K for the condition without a coating layer on the roll surface and 30,000 W/m²K for the Al-coated roll condition. The HTC value of 8000 W/m²K for an uncoated roll surface closely aligns with previously reported values for TRC without a coating layer [23]. This suggests that the average HTC value inferred from the strip temperature is reasonably valid. The increase in the average HTC to 30,000 W/m²K with the coated roll is attributed to improved heat transfer efficiency due to enhanced roll surface roughness and wettability.

Figure 9 presents the simulation results predicting the solidification behavior in the TRC process using the derived HTCs. The results clearly showed that solidification behavior varied significantly depending on the HTC. Furthermore, in the absence of a roll coating layer, residual liquid remained on the strip surface at the roll nip. In this case, since the strip cools slowly under air cooling conditions after exiting the caster, it becomes highly susceptible to hot tearing, as observed in Figure 7a. Therefore, ensuring the formation of an appropriate coating layer on the roll surface to enhance cooling efficiency is crucial for maintaining strip quality.

3.3. Effect of Roll Material on Solidification Behavior in TRC

In this study, Cu rolls were used for the TRC process; however, in mass production, steel rolls are predominantly used due to cost considerations [8]. Therefore, the solidification behavior was predicted under the same conditions, including an Al-coated roll, but with steel rolls instead of Cu rolls. The simulation results are presented in Figure 10. The results showed that when steel rolls, which have lower thermal conductivity than Cu rolls, are used, the cooling efficiency decreased slightly. Consequently, the solid/liquid interface in the sump shifted toward the roll nip direction. Despite this shift, the predicted strip surface temperature and solid fraction at the roll nip were T = 377 °C and fs = 1, respectively, confirming that sufficient cooling was achieved for successful strip production. These results suggest that successful strip production is feasible with steel rolls, provided that a coating layer forms on the roll surface. An additional consideration is the use of a parting agent. In mass production, steel rolls operate under high-load conditions, requiring a parting agent is continuously sprayed onto the roll surface to prevent sticking between the strip and the roll. However, excessive application can reduce interfacial heat transfer efficiency [18]. Therefore, applying an appropriate amount evenly is essential for maintaining stable casting conditions.

3.4. Technical Notes on the Formation of the Al Coating Layer on the Roll Surface

Considering the results thus far, a key technical question arises as follows: How can an Al coating layer be formed on the roll surface during the TRC process? Fundamentally, the formation of an Al coating layer requires a sufficient roll load to the strip. In typical mass production conditions for wide aluminum strips, the casting speed rarely exceeds 1 m/min [8]. Under such conditions, a sufficient roll load is exerted from the early stages of continuous casting, allowing the Al coating layer to form spontaneously as the rolls rotate multiple times. However, if the initial casting speed is too high and the solidified shell is too thin to sufficiently expand the initial roll gap, maintaining such a high speed can hinder the formation of the coating layer of lead to an uneven coating. Figure 11 presents an example of strip surface appearance when the roll surface coating is non-uniform. As shown in the figure, dull regions are irregularly distributed across the strip surface. This dull region corresponds to the part cast under conditions where the coating layer was partially and improperly formed on the roll surface. Such defects cannot be eliminated through subsequent rolling processes, making strict control at the casting stage essential. Therefore, in traditional TRC processes, increasing the casting speed should be approached with caution. To achieve stable strip production at higher speeds, it is beneficial to first establish a stable roll coating layer under conditions where sufficient roll load can be applied before gradually increasing the casting speed.

4. Conclusions

This study examined the influence of roll surface conditions on the solidification behavior and strip surface quality during the twin-roll casting (TRC) process of an Al-5Mg alloy. The principal conclusions are as follows:

• The formation of a stable aluminum coating layer on the roll surface significantly enhanced interfacial heat transfer, leading to improved cooling efficiency and more uniform solidification. The average interfacial heat transfer coefficient (HTC) increased markedly from approximately 8000 W/m²K under uncoated conditions to 30,000 W/m²K when a coating layer was present.
• Experimental observations revealed that strips cast without the coating layer exhibited poor surface quality, characterized by visible hot tearing, surface porosity, and a dull, non-uniform surface appearance.
• In contrast, strips cast under conditions with a uniform Al coating layer showed excellent surface quality, exhibiting smooth, glossy surfaces with no significant surface defects.
• Despite identical process parameters, the final strip thickness differed significantly depending on the roll surface condition. The strip cast without a coating layer had a thickness of approximately 3.0 mm, whereas the strip cast with the coating layer reached 4.8 mm, indicating more effective shell growth and solidification due to improved thermal contact.
• Numerical simulations confirmed that higher HTC values associated with coated roll surfaces facilitated complete solidification before the strip exited the roll nip, thereby reducing the risk of surface defect formation.
• Simulation results also demonstrated the feasibility of employing steel rolls for industrial TRC applications, provided that a stable aluminum coating layer forms to ensure sufficient interfacial heat transfer and solidification.
• The formation and stability of the Al coating layer were found to be highly dependent on the initial roll load. Sufficient roll pressure during the early stages of casting enabled the spontaneous formation of a continuous coating layer, which is critical for maintaining stable casting conditions and achieving high strip surface quality in traditional TRC processes.