Effect of Roll Material on Strip Solidification between the Rolls of a Vertical-Type High-Speed Twin-Roll Caster

Abstract

Tool steel for hot working is usually used for the rolls of conventional twin-roll casters for aluminum alloys. In a vertical high-speed twin-roll caster, copper rolls are used to rapidly solidify the aluminum alloys and to increase the casting speed, since the thermal conductivity of copper is higher than that of tool steel. However, the degree of superiority of copper rolls over steel rolls is not clear. In the present study, the effect of the roll material on strip solidification between the rolls of a vertical high-speed twin-roll caster was investigated by inserting a thermocouple in the strip during casting. The Al-4.8%Mg alloy was used, since it does not stick to the steel rolls in the absence of a parting material. Both copper rolls and mild steel rolls were used. The cooling rate, the supercooling of the liquidus temperature, and the temperature of the strip cast using the copper rolls were superior to those of that cast using the mild steel rolls. The copper rolls were better than the steel rolls in terms of achieving high-speed roll casting, rapid solidification of the strip, and high productivity.

1. Introduction

Twin-roll casters offer advantages over other casting methods, such as shorter processing and rapid solidification. Shorter processing leads to reduced equipment and running costs. Rapid solidification causes a reduction in grain size and in the degree of crystallization at grain boundaries, in addition to an increase in the fraction of the solid solution. Hard brittle alloys, which cannot be formed into plates by ingot casting and rolling, can be cast using twin-roll casters.

In a conventional twin-roll caster for aluminum alloys (CTRCA) [1,2,3,4,5], the rolls are made from tool steel for hot working. The strip is hot-rolled immediately after solidification. The disadvantage of a CTRCA is its low productivity due to the low rolling speed (casting speed). However, if the rolling speed is increased too much, the molten metal does not solidify into a strip, since there is insufficient time for solidification. Since the 1990s, the setback distance and roll load have been increased in order to increase the rolling speed [6,7,8,9,10,11]. A vertical high-speed twin-roll caster (VHSTRC) has also been proposed in order to increase the rolling speed. In the VHSTRC, the rolls are made from copper or copper alloy [12,13,14,15]. The thermal conductivities of copper and tool steel are 398 and 45 W/(m·K), respectively. It is thought that the temperature increase at the roll surface during casting becomes lower as the thermal conductivity of the roll material increases. As a result, molten aluminum alloys can be more rapidly solidified using copper rolls, thus allowing an increase in rolling speed. In a VHSTRC, the roll load is much smaller than in a CTRCA because the copper deformation stress is much smaller than that for tool steel. The deformation stress, especially during hot working, for mild steel is smaller than that for tool steel, but the thermal conductivity of the former material (51 W/(m·K) is larger. In a VHSTRC using copper rolls, no parting material is required, since the aluminum alloy strips do not stick to the rolls. This is in contrast to steel rolls, which generally stick to aluminum alloy strips in the absence of a parting material. The parting material becomes heat resistance, and this suppresses heat transfer between the rolls and the molten aluminum alloy. Exceptions are Al-Mg alloys and Al-SiC_p composites, which do not stick to steel rolls in a VHSTRC. Al-Mg alloys are usually used for sheet forming. For these alloys, the amount of thickness reduction during hot-rolling is smaller than that for pure aluminum and Al-Mn alloys, since Al-Mg alloys are hard. In addition, edge cracking occurs more easily for Al-Mg alloys than for pure aluminum or Al-Mn alloys. Al-SiC_p composites are also hard and brittle, and sheet fabrication from an ingot is usually difficult. The advantage of roll casting is that Al-SiC_p strips can be easily cast using this method [16,17]. In a VHSTRC, the use of mild steel rolls for these alloys offers economic advantages over copper.

The effects of the roll material on strip solidification in a VHSTRC are not clear, particularly in the case of copper rolls, and should be clarified to allow one to make a suitable choice of material. In the present study, the effects of the roll material on the solidification of Al-4.8%Mg strips were investigated using copper rolls and mild steel rolls by inserting a thin K-type thermocouple in the strip during casting [18,19,20].

The tool steel rolls used for a CTRCA usually have a core/shell structure. Cooling water flows between the core and the shell, generally in a circumferential direction. The thickness of the shell is generally about 60 mm. In the VHSTRC used in the present study, the cooling water in the channel flowed in the lateral direction. A channel with a circular cross section was machined in the shell, and the distance between the roll surface and the channel was 6 mm. Such a narrow separation was possible because the roll load in the VHSTRC was small. The mild steel rolls used in this study had a better cooling ability than the steel rolls used in a CTRCA.

The Al-4.8%Mg alloy was used for strip casting, and the strip thickness, the solidification starting point, the temperature at the roll bite, the temperature after exiting the roll bite, the cooling rate, and the degree of supercooling were investigated to clarify the differences between copper rolls and mild steel rolls. The results of the present study show that copper rolls are clearly superior to mild steel rolls for the high-speed casting of most aluminum alloys. However, mild steel rolls are still an economically attractive alternative, depending on the aluminum alloy.

2. Experimental Procedure

The experimental procedure for inserting a thin thermocouple in the VHSTRC is shown in Figure 1. A 0.1 mm diameter K-type thermocouple was used to measure the cooling curves (time–temperature diagrams). A 0.7 mm diameter stainless steel wire (lead wire) was attached to the thermocouple to lead the thermocouple into the molten metal and the strip. The initial roll gap was 1 mm. The roll load was set using the coil springs. The lead wire was set between the rolls rotating at the designated rolling speed, as shown in Figure 1a. The molten metal was poured, and the lead wire was dragged by the strip as shown in Figure 1b. The thermocouple was dragged into the molten metal pool and the strip after a steady state was reached, as shown in Figure 1c. The rolls were pushed by springs, and the roll gap was changed along the thickness of the strip. The tip of the K-type thermocouple, the lead wire, and the thermocouple inserted into the strip are shown in Figure 1d,e, respectively. A GL240-UM-801 data logger (Graphtec Inc., Totsuka, Yokohama, Japan) was used for data recording, and the sampling rate was 100 Hz. The cooling curve was measured using an insulating mold to obtain the liquidus and solidus temperatures at equilibrium. The length, width, and height of the mold were 50, 50, and 40 mm, respectively.

Copper rolls and mild steel rolls were used to investigate the effect of the thermal conductivity of the rolls on the solidification procedure. The thermal conductivity values of mild steel and copper were 51 and 398 W/(m/K), respectively. The diameter and width of the rolls were 300 and 100 mm, respectively. A schematic illustration of the cooling-water channel is shown in Figure 2, and the dimensions of the channel are listed in

Table 1. Dimensions of cooling water channel shown in Figure 1.

Roll Type	Thickness (mm)	Gap (mm)	Diameter (mm)
Copper roll	12	8	8
Mild steel roll	6	7	8

. The rolling speeds were 10 and 30 m/min, and the roll loads were 5 and 20 kN. Al-4.8%Mg was used because the Al-Mg alloy does not stick to the steel rolls even in the absence of a parting material. The chemical composition of the Al-4.8%Mg alloy is shown in

Table 2. Chemical composition of Al-4.8Mg (mass%).

Cu	Si	Mg	Fe	Zn	Mn	Ti	Pb	Sn	Cr	Al
0.03	0.10	4.81	0.18	0.01	0.04	0.02	<0.01	<0.01	0.09	Bal.

. The melt head height was 200 mm. The superheating temperature of the molten metal was 50 °C.

3. Results

3.1. Cooling Curves

Example cooling curves (time–temperature diagrams) for the VHSTRC and the insulating mold are shown in Figure 3. The left and right panes in Figure 3a,b are the cooling curves for the VHSTRC and the mold, respectively. The rolling speed was 30 m/min, and the roll load was 20 kN. The dotted lines indicate the solidus and liquidus temperatures, obtained using the insulating mold. The cooling curves for the VHSTRC had a characteristic shape, where the temperature rapidly dropped and then rose. The minimum temperature following the initial drop (labeled “A” in Figure 3) was the temperature of the roll bite (shown in Figure 1b), based on the time calculated using the rolling speed and the melt head. The temperature of the strip increased after leaving the rolls. The temperatures at the roll bite and after leaving the rolls were lower when copper rolls were used than when mild steel rolls were used.

3.2. Strip Thickness

The effect of the roll material on the strip thickness is shown in Figure 4. The strip thickness decreased as the rolling speed and roll load increased for both types of rolls. The effect of the roll load on the strip thickness decreased as the rolling speed increased. The strip cast using the copper rolls was thicker than that cast using the mild steel rolls. This meant that the productivity was superior using the copper rolls. However, strips thicker than 2 mm could be cast at 30 m/min using the mild steel rolls.

3.3. Strip Temperature at Roll Bite

The effects of the roll material, rolling speed, and roll load on the strip temperature at the roll bite are shown in Figure 5. This temperature is that at “A” in Figure 3. The temperature for the strip cast using the copper rolls was lower than that for the strip cast using the mild steel rolls. In the latter case, the roll load had the largest effect on the strip temperature, followed by the rolling speed. At a rolling speed of 10 m/min and a roll load of 20 kN, the strip temperature was minimized. The strip temperature was the highest at a rolling speed of 10 m/min and a roll load of 5 kN. For the strip cast using the copper rolls, the temperature decreased with increasing rolling speed and roll load. The effect of the roll load on the strip temperature at the roll bite decreased as the rolling speed increased.

3.4. Supercooling

The degree of supercooling of the liquidus temperature, shown in Figure 6, was calculated from the liquidus temperatures obtained from the cooling curves for the VHSTRC and the insulating mold. For the copper rolls, the maximum degree of supercooling was close to 4 °C, whereas that for the mild steel rolls was about 0.7 °C under the same casting conditions. For the copper rolls, the degree of supercooling increased with increasing rolling speed and roll load, but very little change was observed for the mild steel rolls. For both types of rolls, the solidus temperature could not be determined. The temperature at the roll bite was not the solidus temperature, since the strip temperature increased after the roll bite.

3.5. Solidification Starting Point

The starting point for solidification is the region between the roll bite and the position where the liquidus temperature is achieved. The liquidus temperature position was calculated based on the time interval between the liquidus temperature and the roll bite, and on the rolling speed. The effects of the roll material, the rolling speed, and the roll load on the solidification starting point are shown in Figure 7. For both roll materials, the solidification starting point was the shortest at a rolling speed of 10 m/min and a roll load of 5 kN, and the longest at a rolling speed of 30 m/min and a roll load of 20 kN. There was no apparent dependence of the solidification starting point on the roll material used.

3.6. Cooling Rate

The cooling rate was calculated using the times and temperatures at the liquidus temperature and the roll bite. The results are shown in Figure 8, which indicates that the cooling rate was affected by the cooling ability of the roll material. The cooling rate for the copper rolls was larger than that for the mild steel rolls except at a rolling speed of 10 m/min and a roll load of 5 kN. For the strip cast using the copper rolls, the cooling rate increased as the rolling speed and roll load increased. For the strip cast using the mild steel rolls at 30 m/min and 20 kN, the cooling rate was higher than those under other conditions. When the roll load was 20 kN, the cooling rate increased with the increase in rolling speed, similar to the case of the copper rolls. However, when the roll load was 5 kN, the rolling speed had little effect on the cooling rate.

3.7. Strip Temperature after Exiting Roll Bite

The strip temperature increased after exiting the roll bite, as shown in Figure 2. The maximum temperatures after leaving the roll are plotted in Figure 9. At rolling speeds of 10 and 30 m/min, the strips reached their highest temperatures within, respectively, 5 and 2 s after exiting the roll bite. At a rolling speed of 10 m/min and a roll load of 5 kN, the strip temperature was the highest for both roll materials. At a rolling speed of 30 m/min and a roll load of 20 kN, the strip temperature was the lowest for both roll materials. Under the same conditions, a higher maximum temperature was found using the copper rolls. These temperatures were lower than the solidus temperature shown in Figure 3.

4. Discussion

4.1. Cooling Curves

Supercooling occurred, and the liquidus temperature dropped below that in the equilibrium phase diagram. The solidus temperature could not be clearly determined from the cooling curve for the VHSTRC, where the temperature decreased between the rolls, became lowest at the roll bite, and then increased. It is considered that the aluminum alloy around the thermocouple became semisolid at the roll bite as a result of supercooling, and the temperature increased due to the effect of latent heat after exiting the roll bite. The aluminum alloy around the thermocouple did not completely solidify when the strip left the rolls, making the solidus temperature difficult to determine. Clearly, the inside of the strip was semisolid as the strip left the rolls. It is thought that the solidus temperature was lower than the lowest temperature at the roll bite (point A in Figure 3). The temperature of the strip at roll bite A and after leaving the rolls was lower when copper rolls were used. The distance between the water channel and the surface of the mild steel rolls was shorter than that for the copper rolls. However, the temperature of the surface of the copper rolls was lower than that of the mild steel rolls, since the thermal conductivity of copper is higher than that of mild steel. The strip temperature after leaving the copper rolls was lower than that after leaving the mild steel rolls, since the temperature at the roll bite was lower in the former case.

4.2. Strip Thickness

The strips cast using the mild steel rolls were thinner than those cast using the copper rolls. It is thought that the surface temperature of the mild steel rolls was higher than that of the copper rolls, and the heat transfer between the mild steel rolls and the solidified layer in the strip was smaller than that for the copper rolls. As a result, the strip cast using the mild steel rolls was thinner than that cast using the copper rolls. The strip became thinner with the increase in rolling speed and roll load, regardless of the roll material. When the rolling speed increased, the solidification time decreased, and as a result, the strip thickness decreased. The semisolid metal between the solid layers was squeezed near the roll bite in the width direction, forming burrs. The semisolid metal may have been squeezed upward. The volume of squeezed metal became larger with the increase in the roll load. The strip became thinner when the roll load increased, but the effect decreased as the rolling speed increased. The thickness of the strip and the semisolid layer decreased as the rolling speed increased, possibly due to a reduction in the ratio of the thickness of the semisolid layer to that of the strip. As a result, the effect of the roll load on the strip thickness might have decreased as the rolling speed increased. Figure 10 shows the rolling speed dependence of the ratio of the thickness of strips cast using mild steel and copper rolls. The strip thickness decreased more with the decrease in rolling speed using the mild steel rolls than using the copper rolls. This led to a 10% decrease in productivity using the mild steel rolls when the rolling speed was lower than 30 m/min. For the mild steel rolls, a rolling speed of 30 m/min and a roll load of 5 kN, therefore, seem suitable, considering the strip temperature and productivity.

4.3. Strip Temperature at Roll Bite

The temperature of the strips cast using the copper rolls was lower than that of the strips cast using the mild steel rolls, despite the fact that the former strips were thicker. For the mild steel rolls, the temperature at the roll bite was not affected by the rolling speed, i.e., not affected by the strip thickness. For the copper rolls, the temperature at the roll bite decreased with the decrease in strip thickness. This is thought to reflect the heat transfer between the rolls and the solidified layer. The surface temperature for the mild steel rolls was higher than that for the copper rolls, since copper has a higher thermal conductivity, leading to less heat transfer for the mild steel rolls. For the mild steel rolls, the heat transfer was small compared with the thermal conductivity, and the temperature was not affected by the rolling speed (strip thickness).

The effect of the roll material on the relationship between the strip thickness and the strip temperature at the roll bite (Figure 4 and Figure 5) is shown in Figure 11. The difference in strip temperature for the two materials was 50–100 °C, and this increased as the strip thickness decreased. The usually demanded thickness for as-cast strips is 2–5 mm. In this range, the temperature achieved using the copper rolls was more than 50 °C lower than that using the mild steel rolls, indicating the effectiveness of the copper rolls.

4.4. Supercooling

The degree of supercooling was larger when the thermal conductivity of the rolls was higher. The same was reported for mold casting, where the degree of supercooling was larger when the cooling ability of the mold was higher [21]. However, the degree of supercooling of the liquidus temperature for the VHSTRC was lower than that for metal mold casting [21]. The molten metal between the rolls was not stable; it flowed, and it was stirred. This may have been the cause of the lower degree of supercooling of the liquidus temperature. In roll casting, the rolling speed and roll load affect the degree of supercooling. When the rolling speed and roll load increase, the gap between the rolls and the strip thickness decrease. The volume of aluminum alloy between the rolls, therefore, decreases. For these reasons, the degree of supercooling increases.

Figure 5 shows that the degrees of supercooling of the solidus temperature for strips cast using the mild steel and copper rolls were more than 10 and 65 °C, respectively. The strip temperature increased after exiting the roll bite due to the latent heat of semisolid aluminum. This means that the aluminum alloy near the thermocouple had not solidified at the roll bite. The solidus temperature may have been lower than the roll bite temperature. The degree of supercooling of the solidus temperature was much larger than that of the liquidus temperature. The gap between the rolls decreased remarkably as the rolls rotated toward the roll gap. This means that the volume of aluminum alloy rapidly decreased. The aluminum alloy near the thermocouple was semisolid and may have been stable. The cooling rate increased as the volume of aluminum alloy decreased. As a result, the degree of supercooling of the solidus temperature was much larger than that of the liquidus temperature. It is considered that the strip temperature did not decrease until the solidus temperature was reached at the roll bite, as described in Section 4.1. Judging from Figure 5, the solidus temperatures of the strips cast using the mild steel and copper rolls may have been below 540 and 460 °C, respectively.

4.5. Solidification Starting Point

As seen in Figure 7, the difference in the solidification starting point caused by the different casting conditions was within 6 mm. It seems that the starting point was not affected by the thermal conductivity of the roll material as the aluminum alloy rapidly solidified. However, the gap between the rolls decreased with the decrease in thermal conductivity because the strip thickness decreased and the roll gap narrowed. As a result, the volume of aluminum alloy between the rolls decreased, and the solidification starting point was not greatly affected by the thermal conductivity (cooling ability) of the rolls.

4.6. Cooling Rate

As the rolling speed increased, the volume of aluminum alloy between the rolls rapidly decreased, because the strip became thinner and the roll gap became narrow. This means that the amount of heat decreased. The distance between the roll surface and the center of the strip, which became heat resistant, decreased. Thus, the alloy cooled much more quickly. As the roll load increased, the roll gap narrowed, and the volume of the aluminum alloy decreased. The increase in the roll load increased the heat transfer between the solidified layer and the roll surface because of an increase in the contact area. Thus, for the copper rolls, the cooling rate increased as the rolling speed and roll load increased, as shown in Figure 8. When the mild steel rolls were used, the heat flow from the solidified layer to the rolls was smaller than that for the copper rolls, because the surface temperature of the mild steel rolls was higher than that of the copper rolls. This might have been due to the lower cooling rate and the weaker effects of the rolling speed and roll load (strip thickness) on the cooling rate in the case of the mild steel rolls.

4.7. Strip Temperature after Exiting Roll Bite

The temperature of the strips cast using the copper rolls was lower than that of strips cast using the mild steel rolls even though the former strips were thicker, as shown in Figure 9. The increase in the strip temperature after leaving the roll was caused by the latent heat of the partially solidified aluminum alloy inside the strip. The rate of increase in strip temperature decreased as the volume of non-solidified aluminum alloy decreased. Therefore, the strip temperature after leaving the rolls decreased under conditions where the strip temperature at the roll bite and the strip thickness decreased.

5. Conclusions

Cooling curves for solidifying aluminum alloy strips cast using a VHSTRC were obtained by inserting a thin K-type thermocouple in the strips, and the effects of the rolling speed, roll load, and roll material on the solidification process were investigated. The following conclusions were obtained:

(1)

The strip thickness increased with the increase in the thermal conductivity of the roll and the decrease in rolling speed and roll load;

(2)

The distance from the roll bite to the solidification starting point ranged from 37 to 43 mm and increased with the decrease in strip thickness;

(3)

The liquidus temperature could be determined, but the solidus temperature could not be determined;

(4)

The degree of supercooling and the cooling rate increased with the increase in the thermal conductivity of the roll and the increase in rolling speed and roll load;

(5)

The strip temperature increased after leaving the roll, presumably because the strip interior was not solidified;

(6)

The strip temperature at the roll bite and after leaving the roll decreased with the increase in the thermal conductivity of the roll, rolling speed, and roll load.

For the mild steel roll used in this study, the distance between the roll surface and the cooling water channel was 6 mm, which is about one-tenth of that for a CTRCA. This was realized by machining the channels in the shell of the rolls. The cooling ability of these mild steel rolls may be better than that of rolls used in a CTRCA.

Al-4.8%Mg could be continuously cast into strips at 30 m/min using the mild steel rolls in the present study, without any sticking occurring. When the solidification length (Figure 1a) was 100 mm, a rolling speed of 30 m/min may have been the maximum, considering the strip temperature at the roll bite. The solidification length could be increased using a larger-diameter roll. The casting speed may be increased using a greater solidification length. In a CTRCA, the roll diameter is usually 600–1200 mm. The fabrication of a large-diameter roll using mild steel may not be difficult. The use of mild steel offers a more economical solution, since machining and fabrication are easier than for copper. In addition, the roll life may be longer for mild steel, since the deformation stress is larger than that for copper.

However, copper rolls were shown to be highly effective for high-speed roll casting. The internal temperature at the roll bite for strips cast using copper rolls was more than 50 °C lower than that using mild steel rolls, and the surface temperature may have been even lower. The cooling rate of strips cast using copper rolls was higher than that obtained using mild steel rolls. The solid solubility limit is also higher when copper is used. In addition, copper rolls can be used for most aluminum alloys without the need for a parting material. Mild steel rolls can only be used for Al-Mg alloys and Al-SiC_p composites without a parting material; however, Al-Mg strips are in great demand.

The results of the present study show that copper rolls are clearly superior to mild steel rolls for the high-speed casting of most aluminum alloys. However, mild steel rolls are still an economically attractive alternative, depending on the aluminum alloy.

Thermal conductivity is an important factor for determining the most appropriate roll material for high-speed twin-roll casters to realize rapid solidification. Due to their high thermal conductivity compared with steel, copper and its alloys are the best choices for the roll material. Even using mild steel with a very thin 6 mm thick shell, a suitable cooling ability cannot be achieved.