2017 Volume 58 Issue 6 Pages 967-970
In order to develop a numerical simulation model of the high-speed twin-roll casting (HSTRC) process, an appropriate melt/strip heat transfer coefficient (HTC) is required. In the present study, the strip temperature was directly measured by using the traveling thermocouple technique. Then, we tried to determine the reasonable HTC values, which conformed well with the temperature curves obtained experimentally. The results showed that the HTC values increased with increasing casting speed. Moreover, the overall HTCs were much higher than those reported so far. We believe that a large hydrostatic pressure, induced by a high melt pool level, and a sufficient roll separating force helped improve the contact condition between the solidifying shell and the roll surface.
Twin-roll casting is a well-known cost-effective process for producing aluminum sheets. Using this process, 3–10 mm-thick thin strips can be fabricated directly from the molten metal. However, because aluminum alloys have relatively poor high-temperature stiffness, horizontal-type casters are used, which are operated at a low casting speed (typically lower than 2 m/min in mass production). Recently, Haga's group at Osaka Institute of Technology, Japan, has developed a vertical-type caster with a remarkably increased casting speed (up to 180 m/min).1) Many studies have conducted trials to fabricate various types of aluminum alloy strips via vertical-type high-speed twin-roll casting (HSTRC).2–5) However, several problems are encountered during the casting process, such as surface cracking, inverse segregation, and internal cracking, especially with alloys having a wide freezing temperature range.3,5) Internal cracking is a severe defect that forms because of the liquid remaining at the central region of the strips when they pass through the roll nip.5) In order to control the internal crack effectively, we need to understand the cooling behavior of the strips systematically via experiments and numerical simulations.
In order to develop a numerical simulation model of HSTRC, an appropriate heat transfer coefficient (HTC), hmelt/roll, at the interface between the molten metal and the casting roll is required. In the present study, the temperature of the central region of the strips was directly measured by the traveling thermocouple technique. Then, the temperature data were used to determine the HTC. Based on the results, we discuss the relationship between the HTC and the casting conditions.
Al-2 mass%Si alloys were fabricated by melting pure Al (99.9 mass%) and Al-Si ingots in an electric furnace. Then, a degassing process was conducted using Ar gas for 10 min, prior to the HSTRC. A schematic illustration of the vertical-type HSTRC is shown in Fig. 1 (a); a detailed description of the HSTRC procedure can be found elsewhere.5) The initial roll gap was 1 mm. The roll separating force (RSF) was set to be ~60 kN during the HSTRC. In order to measure the strip temperature, a specially designed K-type thermocouple, connected with a dummy line, was flown directly into the molten metal pool during the casting. The moving thermocouples were set to be finally positioned at a steady-state part of the continuous cast strip where the strip thickness was constant.5) The temperature was measured from the center line of the upper surface of melt pool to the strip center (Fig. 1 (b)). Melt temperature was kept constant in the feeding nozzle part (ⓐ region in Fig. 1 (b)), and then it was dropped rapidly as the thermocouple was sandwiched between the two solidifying shells growing on both the roll surfaces (ⓑ region in Fig. 1 (b)). For each casting trial, we confirmed the position of the thermocouple, i.e., we checked whether the thermocouple was inserted well into the center of the cast strips or not (Fig. 1 (c)).
(a) Schematic illustration of HSTRC with a traveling thermocouple; (b) temperature changes measured by using the traveling thermocouple; (c) transverse cross-section of the as-cast strip revealing the embedded thermocouple.
Heat transfer simulations were conducted to investigate the temperature distribution during the HSTRC. The temperature data obtained by the traveling thermocouple technique were applied to determine the HTC at the interface between the molten metal and the casting roll. The model geometry and boundary conditions are given in Table 1. These numerical simulations were carried out using the commercial ProCAST software. CompuTherm database was used for thermo-physical data. The calculation was conducted under the Scheil cooling condition considering very high cooling rates (1000–10000℃/s) of the HSTRC process.3)
Casting alloy: Al-2 mass%Si Initial melt temperature: 655℃ Liquidus temperature : 648℃ Solidus temperature : 575℃ |
||||
Temperature (℃) |
Thermal conductivity (W/mK) |
Density (kg/m3) |
Enthalpy (kJ/kg) |
|
306 | 191 | 2635 | 270 | |
576 | 193 | 2570 | 565 | |
648 | 92 | 2409 | 1036 | |
800 | 92 | 2358 | 1235 | |
Interfacial heat transfer coefficient (HTC, h) hmelt/nozzle: adiabatic hstrip/air: 12 W/m2K6) hmelt/roll: variable Roll temperature: 50℃7) Air temperature: 25℃ |
HSTRCs were carried out at various casting speeds of 30–90 m/min (0.5–1.5 m/s). Figure 2 shows experimental and simulation results of temperature changes at the strip center near the roll nip. The resultant strip thickness used in the present simulation changed according to the experimentally obtained values. At 30, 60, and 90 m/min, the thicknesses were 4.4, 3.0, and 2.6 mm, respectively. For each condition, numerical simulations were conducted by changing the HTC to determine the appropriate value. The results showed that the HTC values, at which simulation results reasonably conformed with the experimental works, changed with the changing casting speed. The reasonable HTC values for 30, 60, and 90 m/min casting speeds were 32000, 40000, and 60000 W/m2K, respectively.
Temperature change in the strip center at various casting speeds: (a) 30 m/min, (b) 60 m/min, and (c) 90 m/min.
Wang and Matthys8) have reported a comprehensive result: the HTC was closely related with the substrate moving velocity in various casting methods such as twin-belt, twin-roll single-roll, fee-jet melt spinning, and planar flow casting. The HTC values increased with substrate moving velocity, regardless of the casting methods. Using these data, Wang and Matthys developed the following empirical relation8):
\[ \bar{h} = 17.5v_{s}^{0.65} \] | (1) |
The fact that the HTC increases with increasing casting speed is also applicable to the present study. Figure 3 shows the relationship between the casting speed and the HTC, obtained in the present study and by using eq. (1). The HTC values increases with increasing casting speed, showing a similar trend with the data obtained using eq. (1). However, at the same casting speed, our values are much higher than those calculated by eq. (1). Hence, we proposed the following empirical relation:
\[ \bar{h} = 44.8v_{s}^{0.55} \] | (2) |
Relationship between the casting speed and the averaged HTC in HSTRC.
Two reasons can be given to explain the higher HTC values than those reported by Wang and Matthys8). The first is increased hydrostatic pressure in HSTRC. Generally in a vertical-type twin-roll caster for steel, the feeding nozzle for supplying the molten metal is directly submerged into the melt pool; therefore, meniscus contacting the roll surface can be exposed to the air or shield gas. This means that there is a limit up to which hydrostatic pressure can increase during casting. Most of the other casting methods mentioned above like single roll casting, and planar flow casting also suffer from geometrically limit that hinders the large hydrostatic pressure from becoming too large. In contrast, in the caster used in the present study, a large feeding nozzle that directly mounted on the roll surface helped build a high melt pool head (Fig. 1 (a)). This can provide a relatively large and stable hydrostatic pressure during the HSTRC. It has been shown that using enough melt pool head is effective in improving the contact condition between the solidifying shell and the casting roll.5) Increased strip thickness and better surface condition are indirect evidence of the improved contact condition.
The second factor is the RSF. It has been earlier shown that cooling rates at the roll nip increased sharply as the RSF increased, but it remained constant above the critical RSF.3) This indicates that the formation of air gap between the rolls and solidifying shells induced by solidification shrinkage can be reduced to some extent by the RSF. These factors are considered to be reasonable for the overall increased HTC in the present HSTRC.
The improved contact condition between the strip and the casting rolls, characterized here by the increased HTC, is closely related with increased cooling rate of strip. For the HSTRC of aluminum alloys, the strips should be cooled down sufficiently below the solidus temperature when they pass through the roll gap. Otherwise, the hot strips could be torn down easily due to their poor high-temperature stiffness. Moreover, if some liquid is remained when the strips come out from the roll gap, it can cause severe internal cracking at the central region of the strip.5) Therefore, the improved contact condition, i.e. the increased cooling ability under the HSTRC condition used in the present study is very effective to maintain the high productivity and quality of strips.
In order to obtain an appropriate melt/strip HTC for the simulation, the traveling thermocouple technique was adopted to measure the strip temperature directly during the HSTRC. HTC increased with increasing casting speed. Moreover, compared to the literature, the present study showed an overall higher HTC, proposing a new empirical relation, $\bar{h} = 44.8v_{s}^{0.55}$. This indicates that our casting conditions (with sufficient hydrostatic pressure of melt and roll separating force) are effective in obtaining a much improved contact condition between the strip and the casting rolls, which is closely connected with improved productivity and strip quality.
This work was supported by the Main Research Program of Korea Institute of Materials science (KIMS), Republic of Korea.