ISIJ International
Online ISSN : 1347-5460
Print ISSN : 0915-1559
ISSN-L : 0915-1559
Casting and Solidification
Effect of Shear Stress on Heat Transfer Behavior of Non-Newtonian Mold Fluxes for Peritectic Steels Slab Casting
Shaopeng GuGuanghua WenJunli GuoZhe WangPing TangQiang Liu
Author information
JOURNALS OPEN ACCESS FULL-TEXT HTML

2020 Volume 60 Issue 6 Pages 1179-1187

Details
Abstract

A new mold flux based on non-Newtonian fluid for the peritectic steels casting was prepared. The heat transfer behavior and lubrication property of this non-Newtonian mold flux were examined by heat flux simulator, ultraviolet-visible-near-infrared spectrometer (UV-Vis/NIR), Raman spectroscopy, SEM, and confocal laser scanning microscopy (CLSM), and the results were compared with the conventional mold flux used in peritectic steels casting. The results showed that the heat transfer property of liquid layer of N1 slag was reduced through the destruction of silicate network structure by shear stress. Compared with the data obtained under static and stirring conditions, the qmax and degree of polymerization (DOP) of N1 slag were reduced from 0.921 MW/m2, 0.728 to 0.716 MW/m2, 0.583, respectively. However, the shear stress has no effect on the heat transfer property of liquid layer of N0 slag. Second, the heat transfer properties of solid slag layer of N0 and N1 slag were all inhibited through increasing the crystallization rate, crystallization fraction, and slag film thickness by shear stress. While, under stirring condition, the slag film thickness and t2 of N1 slag was lower than that of N0 slag. Third, the heat transfer behavior of air gap layer of N0 and N1 slag were all controlled by shear stress. The surface roughness (Ra) and shedding time of N0 and N1 slag with agitation were increased to 54.49 um, 61 s and 52.87 um, 59 s, respectively. Finally, the break temperature of N1 slag was 9 K lower than that of N0 slag.

1. Introduction

Mold fluxes are vital functional materials and play an indispensable role in the continuous casting of steels.1,2) They perform the following functions: (1) protecting steel surface from oxidation, (2) providing thermal insulation to prevent steel from freezing, (3) absorbing non-metallic inclusions, (4) providing an optimal level of horizontal heat transfer, and (5) providing liquid flux to lubricate the newly formed steel shells. Among them, heat transfer performance is one of the most important determiners of the quality of peritectic steels obtained by a continuous casting technology.3)

In peritectic-steel continuous casting process, longitudinal cracks have always been a serious problem influencing the quality of steel shells, which is mainly caused by a volumetric shrinkage occurring during δγ phase transformation.4) It can be relieved with an application of a “mild cooling” strategy in a mold, especially in the initial stage of solidification.5) Presently, the main method employed in steel industries is to improve the basicity of mold fluxes to realize a “mild cooling” strategy. Because the crystallization property can be improved by these high-basicity mold fluxes. Concurrently, it will lead to a larger thickness of a solid slag film. However, the lubrication of steel shells will be deteriorated by these high-basicity mold fluxes. Therefore, a new mold flux need to be developed to resolve this contradiction problem. Such as a mold flux with low-basicity and high heat transfer control property. Yoon et al.6) indicated that the radiative heat transfer could be controlled by adding metallic iron to mold fluxes, which could make the “low-basicity and high heat transfer control mold fluxes” came true. Some researchers7,8,9,10) prepared mold fluxes with low basicity and high heat transfer control property by an addition of transition metal oxides, such as Fe2O3, FeO, MnO, MoO3, and ZrO2. Yang et al.11) studied bubbles on heat transfer behavior of mold fluxes, and attempted to moderate the bubble amount and size in molten slag for controlling the heat transfer behavior. Tsutsumi et al.12) studied the relationship between cooling rate and surface roughness of solid slag film and provide fundamental information on the mechanism of decreasing the heat transfer from steel shells to mold. Yamauchi et al.13) deeply studied the mechanism of heat transfer from the steel shells to mold and indicated that the large air gap between mold and solid slag film and small size of silicate ions in molten slag could inhibit the heat transfer in a mold. In conclusion, more efforts had been made to solve the contradiction problem in casting of peritectic steels. However, this contradiction has not be solved perfectly and more studies are still needed.

In this work, a non-Newtonian mold flux with shear thinning behavior under shear stress is proposed to settle the above problem. During a continuous casting process, mold fluxes are subjected to shear stress caused by mold oscillation and casting slab movement,14) as shown in Fig. 1, and the property of mold fluxes is seriously affected by shear stress. Shin et al.15) prepared a non-Newtonian mold flux, which was based on the fact that the shear stress exerted on mold fluxes was similar to that of non-Newtonian fluid, and studied the effect of shear stress on its rheological property. However, the heat transfer property of this non-Newtonian mold fluxes under shear stress did not be evaluated. Watanabe et al.16) conducted a laboratory continuous casting test with non-Newtonian mold fluxes and found that the lubrication and heat transfer performances were improved. Nevertheless, the basicity of this non-Newtonian mold fluxes is too low to form crystals. Since the cracks originate from the meniscus area (the occurrence of peritectic reaction), and large amount of liquid layer, small amount of solid layer (including a glass layer and a crystal layer) and air gap layer are included in this area, heat transfer property of each layer at the initial stage of solidification is very important to the quality of peritectic steels. However, to the best of our knowledge, the effect of shear stress on heat-transfer performance of each layer of the non-Newtonian mold fluxes has not been reported yet, let alone make the application of non-Newtonian slags for the peritectic steels casting process. Therefore, in view of the need for “low-basicity and high heat transfer control mold fluxes”, it is very necessary to understand this influence deeply and systematically, which can provide a theoretical basis for developing non-Newtonian mold fluxes that can be applied in a continuous casting of peritectic steels.

Fig. 1.

Schematic diagram of shear stress suffered by mold flux and mold flux. (Online version in color.)

In this study, a non-Newtonian mold flux and a conventional mold flux applied for the peritectic steels casting were prepared. The effect of shear stress on heat transfer performance of liquid layer, solid layer (glass layer and crystal layer) and air gap layer (interface thermal resistance) in these slags were characterized by heat flux simulator, ultraviolet-visible-near-infrared spectrometer (UV-Vis/NIR), scanning electron microscope (SEM) and confocal laser scanning microscopy (CLSM). The structure evolution of these slags under shear stress were visualized by Raman spectroscopy. Finally, the lubrication performance of these slags were evaluated.

2. Experimental Procedure

2.1. Material Preparation

Reagent-grade powders of CaO, SiO2, MgO, Al2O3, Na2CO3, CaF2, and Si3N4 were used as the raw materials and the composition of mold flux before experiment and after experiment were given in Table 1. The chemical component of each mold flux after experiment was analyzed by X-ray fluorescence spectroscopy (XRF). It could be found that the content of each chemical component did not change obviously. Among them, N0 slag is a conventional mold flux for the peritectic steels casting.15) N1 slag is a non-Newtonian mold flux. According to the proportion in Table 1, one 250 g and two 350 g samples were prepared, mixed, and calcined in a muffle furnace at 373 K to remove moisture. The 250 g sample was placed in a silicon-molybdenum electric furnace at 1573 K for 20 min to ensure complete melting. Then, a viscosity-temperature curve was obtained using a viscometer. One of the 350 g samples was placed in a HF-200 heat flux simulator (it was developed by Professor Wen at Chongqing University and was used to quantitatively characterize the heat transfer behavior of mold fluxes. The detailed description could be seen in References 15 and 16) and maintained at 1673 K for 20 min to homogenize its chemical composition. Then, a stirrer was adopted to molten slag for 5 min at 200 rpm. After that, A preheated steel bar (Φ15 mm × 80 mm) was dipped into the molten slag for 10 s and quickly extracted for quenching and then drying (the samples obtained were defined as the “stirring high-temperature samples”). The remaining molten slag was used for heat flux curve calculation. The above process was repeated but ignored the agitation process, which was a comparison with the above test. Then, the samples and heat flow curves were obtained, and the samples obtained were defined as the “static high-temperature samples”. The other 350 g sample was first heated to 1673 K for 20 min and then decreased to 1471 K for 20 min in HF-200 heat flux simulator. After that, the above procedures were repeated again including the stirring and static experiments, and the slag films (defined as the “stirring low-temperature samples” and the “static low-temperature samples”) and heat flux curves were obtained. The stirring high-temperature samples and static high-temperature samples were examined using UV-VIS-NIR and Raman spectroscopy. The stirring low-temperature samples and static low-temperature samples were examined by UV-VIS-NIR, SEM, DSC, and CLSM.

Table 1. Chemical compositions of mold flux (wt%).
Before experiment (wt%)After experiment (wt%)
Weight amount using balanceXRF analysis Calculated
SampleCaOSiO2MgOAl2O3Na2OFNCaOSiO2MgOAl2O3Na2OFN
N045.933.92.03.08.07.2045.9233.942.083.037.977.010
N142.536.32.03.08.07.2142.5336.312.043.027.957.040.98

2.2. Experimental Methods

UV-VIS-NIR spectrometer was used to test the transmissivity and reflectivity of all the samples in the wavelength range 300–2300 nm.

Heat flux simulator17) was adopted to obtain the heat flux curves of samples. The schematic of the experimental apparatus was shown in Fig. 2. A water-cooling copper sensor (20 mm × 25 mm × 35 mm) was used to simulate a mold. During the experiment, 350 g mold flux was heated to 1673 K for 30 min in HF-200 heat flux simulator and then a stirring process at required rotational speed was conducted for 5 min. After that, the copper sensor was dipped into the molten slag. A solid slag film was formed around the copper sensor. The heat across the slag film was removed by cooling water (the flow of cooling water was maintained at 200 L/h to ensure that the flow of cooling water per unit area was similar to that in a real mold) and the temperature difference of in/out cooling water was recorded by a computer. After 45 s, the copper sensor wrapped with solid slag film was extracted from the molten slag and the time from the extraction of solid slag film from molten slags to the drop of solid slag film from copper sensor was recorded, which is defined as the “shedding time”. The heat flux density can be calculated through Eq. (1):   

q=WC (Ta-Tb) / (1   000F) (1)
where, q is the heat flux density through a solid slag layer (MW·m−2), W is the flow rate of cooling water (kg·s−1), Ta and Tb are the temperatures of water in the inlet and outlet, respectively (K), F is the effective surface area of the copper sensor (m2), and C is the heat capacity of water (kJ·(kg·K)−1).
Fig. 2.

Schematic of experimental apparatus and slag films obtained by this apparatus. (Online version in color.)

A typical heat flux curve of a mold flux is shown in Fig. 3. The measured heat flux was plotted against testing time. The regression function from time t1 to 45 s follows Eq. (2):   

q=A t -B (2)
According to Eq. (2), the time at which the slope of curve is −1 is defined as the “characteristic time”, as shown in Fig. 3. The equation of t2 follows:   
q , =-AB t -(B+1) (3)
The characteristic time t2 can be expressed as:18)   
t 2 =exp[ lnAB/(B+1) ] (4)
where, q is the instantaneous heat flux of the slag (MW·m−2), t is the immersion time of the copper sensor (s), and A and B are regression coefficients.
Fig. 3.

Typical heat flux curve of samples. (Online version in color.)

The variation in heat flux density with time was classified into three stages.19) In the initial stage (0-t1), heat flux density increased quickly and reached a maximum value (qmax) at t1. In this stage, copper sensor was firstly immersed into liquid slags and was wrapped by high-temperature molten slags. The heat moved from liquid slag film to copper sensor, and it was gradually increased until the maximum value. Due to the thickness of solid slag film being tiny in this stage, the heat resistance of liquid slag is the major contributor to heat transfer, which means that the heat-transfer performance of liquid slag film can be represented by qmax.20) During the second stage (t1-t2), the heat flux density sharply decreased with increasing immersion time until the characteristic time t2, indicating a formation of solid layer (including a crystal layer and a glass layer) around the copper sensor. Concurrently, the thickness and crystal fraction of solid layer increased with immersion time. t2 is the turning point between the sharp decrease and the slow decrease of heat flux density, indicating a completed formation of air gap layer. Therefore, the heat-transfer performance of solid slag film and air gap layer can be represented by the heat flux density at t2 (qt2). Many studies21,22) had shown that interfacial thermal resistance was caused by the generation of crystals. Therefore, the crystallization rate of mold fluxes can be characterized by the magnitude of t2 to a certain extent. In the last stage (t2-45 s), heat flux decreased slowly with time because only a slight increase of the thickness of solid slag film was conducted.

The viscosity-temperature curve of mold fluxes was obtained by a rotating cylinder method at a cooling rate of 6 K/min, and the break-temperature was obtained from the viscosity-temperature curve by a tangential method.23)

The structure of amorphous samples was analyzed using Raman spectroscopy at room temperature. The principle of Lab-RAM had been described in detail in previous papers.24) The acquired Raman spectra was curve-fitted with Gaussian functions using Origin Microcal software. The area of the deconvoluted peaks was used to calculate the degree of polymerization (DOP) of samples.

SEM, DSC, and CLSM were adopted to analyze the morphology of crystals, crystallization fraction, and the surface roughness (Ra) of the low-temperature slag films obtained at static and stirring conditions, respectively.

3. Results and Discussion

3.1. Heat-transfer Property of Liquid Slag Film

3.1.1. Radiative Heat Transfer of Liquid Slag

The radiative heat transfer across mold fluxes contributes to 20%–50% of the total heat transfer in a mold.7) Therefore, during the casting of crack-sensitive steel grades, it is very important to reduce the radiative heat transfer. Especially the radiative heat transfer of mold flux in the initial stage of solidification, where a large amount of liquid layer and a few crystals were existed. Figures 4(a) and 4(b) show the transmissivity and reflectivity curves of the stirring high-temperature samples and static high-temperature samples obtaining from the N0 and N1 slags, respectively. As shown, in non-Newtonian slag (N1), the transmissivity of the stirring high-temperature sample was lower than that of the static high-temperature sample at the wavelengths of 300–2300 nm. The reflectivity of the stirring high-temperature sample was greater than that of the static high-temperature sample at wavelengths of 300–2300 nm. However, in N0 slag, the transmissivity and reflectivity of the stirring high-temperature sample and the static high-temperature sample at the wavelengths of 300–2300 nm were almost the same. This may be caused by the change of the structure of non-Newtonian slag.25) Radiative heat transfer was generally formed through the photon transfer; however, the photon mean free path of the stirring high-temperature sample of N1 slag decreased due to the stirring process. Therefore, the radiative heat transfer decreased. Diao et al.26,27) indicated that infrared radiation absorption was enhanced when the symmetrical characteristics and dipole moments of molecular structures were modified during vibration. The more the dipole moment modification in the molecular vibrations, the more infrared radiation absorption. As the N atom was participated in the silicate network structure by replacing the non-bridged oxygen ions and the weak N-M1+/2+ (M was the alkali or alkaline earth metal ions) bonds were formed in the non-Newtonian slag, the structure of non-Newtonian slag was easy to be broken by agitation, which ultimately increased the radiation absorption. Therefore, the transmissivity was lower in the stirring high-temperature sample of the non-Newtonian slag and the reflectivity was higher in the stirring high-temperature sample of the non-Newtonian slag. However, an in-depth mechanism investigation was not possible in the present study and will be examined in the future. For the N0 slag, since the agitation behavior has no effect on the structure of molten slag, the transmissivity and reflectivity of N0 slag with and with not stirring were almost the same.

Fig. 4.

Optical properties of high-temperature samples of N0 and N1 slag under static and stirring condition: (a) transmissivity curves and (b) reflectivity curves. (Online version in color.)

3.1.2. Heat Flux Density of Liquid Slag

The qmax of the stirring high-temperature samples and static high-temperature samples obtaining from the N0 and N1 slags are summarized in Table 2. As shown, for the N1 slag, the qmax of the static high-temperature sample was larger than that of the stirring high-temperature sample. The heat flux density at 1673 K was affected by both the radiative heat transfer and the conductive heat transfer, and the radiative heat transfer accounts for a large proportion. Figure 4 has shown that the radiative heat transfer reduced in N1 slag due to the agitation process. The conductive heat transfer was formed by phonon conduction and was directly proportional to the phonon mean free path. The phonon mean free path of N1 slag was reduced by the agitation process.28,29) Hence, the conductive heat transfer of the stirring high-temperature sample of N1 slag was also decreased. Ultimately, for N1 slag, the qmax of the stirring high-temperature sample was lower than that of the static high-temperature sample. For the N0 slag, the qmax of the static high-temperature sample and the stirring high-temperature sample was almost the same.

Table 2. The qmax of static high-temperature and stirring high-temperature samples obtained from N0 and N1 slags.
Samplesqmax (MW/m2)
N0Static high-temperature slag0.903
Stirring high-temperature slag0.898
N1Static high-temperature slag0.921
Stirring high-temperature slag0.716

3.1.3. Structure of Molten Slag

The Raman spectral fragments between 400 and 1400 cm−1 of the high-temperature samples of N0 and N1 slags obtained under static and stirring conditions are plotted in Fig. 5(a). Raman shifts ranging from 800 to 1200 cm−1 were associated with the Si–O stretching vibrations of the mostly depolymerized silicate discrete anions (Qn units, where n is the number of bridging oxygen atoms (n = 0–4)).30,31,32) The characteristic peaks in the Raman spectra of the various structural units were obtained by typical fitting until an R2 value >99.5% was achieved, as listed in Table 3. The typical deconvolution result for mold fluxes is shown in Fig. 5(b). The bands in the ranges 850–880, 900–920, 950–1000, and 1050–1100 cm−1 corresponded to Q0, Q1, Q2, and Q3 units, respectively. Due to the Q4 unit (1060–1200 cm−1) was extremely difficult to be observed in the mixed silicate system, the Q3/Q2 ratio was regarded as the DOP of molten slags according to the following equilibrium reaction among the silicate units:15)   

Si 2 O 5 = SiO 3 + SiO 2 (5)
  
K= [SiO 3 ] [SiO 2 ] [Si 2 O 5 ] = Q 2 Q 4 Q 3 (6)
  
Q 4 =K Q 3 Q 2 degree   of   polymerization (7)
where, K is the equilibrium constant of Eq. (5). Hence, Q3/Q2 is directly proportional to Q4, the concentration of the severely polymerized unit.
Fig. 5.

Raman spectra: (a) Raman spectra data for each high-temperature samples; (b) Typical deconvolution of Raman spectra for mold flux. (Online version in color.)

Table 3. Peak analysis of Raman spectra in mold flux.
Static sampleDynamic sampleRaman bandsRe.
Raman shift (cm−1)672683Bending Si–O vibration of [SiO3]2−[8]
865869SiO44− stretching in monomer structure unit (Q0)[20]
902910SiO76− stretching in dimer structure unit (Q1)[27]
956966SiO32− stretching in chain structure unit (Q2)[28]
10491052Si2O5 stretching in sheet-like structure unit (Q3)[29]

The integrated area of the individual deconvoluted peaks provided a semi-quantitative evaluation of the number of Qn units. The peak area ratios and DOPs of the static high-temperature samples and stirring high-temperature samples obtained from N0 and N1 slags are listed in Table 4. For the N1 slag, Q1 and Q2 in the stirring high-temperature samples were higher than that of the static high-temperature samples. Conversely, Q0 and Q3 in the stirring high-temperature samples were lower than that of the static high-temperature samples. Accordingly, DOP of the stirring high-temperature samples decreased. It was because that the structure of non-Newtonian slag was related to the shear stress produced by an agitation process.15) For the N0 slag, there was no obvious change of DOP for each high-temperature samples.

Table 4. Polymerization degree and relative content of the high-temperature samples of N0 and N1slags under static and stirring conditions according to Raman peaks.
SamplesPolymerizationBest-fitted Gaussian peak area ratio
Q0Q1Q2Q3
N0Static high-temperature sample0.5740.1270.1610.4540.261
Stirring high-temperature sample0.5680.1210.1650.4570.259
N1Static high-temperature sample0.7280.1240.1310.4340.316
Stirring high-temperature sample0.5830.1120.1710.4610.269

3.2. Heat-transfer Property of Solid Slag Film

3.2.1. Break Temperature

The break temperature of mold fluxes was obtained by a tangential method, and 1471 K was selected as the temperature for studying the heat transfer of solid layer. Mills33) pointed out that for the crystalline mold flux, crystals first precipitated in molten slags at the break temperature.

3.2.2. Radiative Heat Transfer of Solid Slag Film

Figures 6(a) and 6(b) show the transmissivity and reflectivity curves of the stirring low-temperature samples and static low-temperature samples obtaining from the N0 and N1 slags, respectively. For the N1 slag, the transmissivity of the stirring low-temperature samples was lower than that of the static low-temperature samples. The reflectivity of the stirring low-temperature samples was higher than that of the static low-temperature samples. For the N0 slag, the transmissivity and the reflectivity had the same trend as N1 slag. Radiative heat transfer of solid slag films was strongly affected by the crystalline fraction and slag film thickness.1,27) The transmissivity decreased and reflectivity increased with the increase of crystalline fraction and slag film thickness. The crystallization ability of molten slags was improved by an agitation process, and the further mechanism will be discussed in detail in the next section. Comparing the transmissivity and reflectivity results of the N0 and N1 slags, for the static low-temperature samples, the transmissivity of N0 slag was lower than that of N1 slag, and the reflectivity of N0 slag was larger than that of N1 slag. It was because that the original basicity of N0 slag was larger than that of N1 slag, which resulted the stronger crystallization ability of N0 slag. For the stirring samples, the transmissivity and reflectivity results of the N0 and N1 slags are similar in value.

Fig. 6.

Optical properties of low-temperature samples of N0 and N1 slags under stirring and static conditions: (a) transmissivity curves and (b) reflectivity curves. (Online version in color.)

3.2.3. qt2, Crystallization Rate, Crystalline Fraction, and Slag Film Thickness

The qt2, crystallization fraction, and slag film thickness of each low-temperature samples obtaining from the N0 and N1 slags are listed in Table 5. t2 is the characteristic time of each high-temperature samples obtaining from the N0 and N1 slags, which represents the rate of the precipitation of crystals from the liquid slag film. According to Table 5, for the N0 slag, the qt2 and t2 of the stirring samples were smaller than that of the static samples. The crystallization fraction and slag film thickness of the stirring low-temperature samples were greater than that of the static low-temperature samples. For the N1 slag, the qt2, t2, crystallization fraction, and the slag film thickness of the low-temperature samples obtained under stirring and static conditions had the same trend as that of the N0 slag. The number of secondary nucleation sites and the kinetic energy of the ions involved in crystallization were increased by a shear stress. Meanwhile, the structure of the formed crystal in stirring samples might be incomplete, leading to more defects.28) In addition, the Q1 unit in N1 slag under stirring condition was increased, and Q1 was the dominant structure unit in the crystal structure of cuspidine.34) Comparing the qt2, t2, crystallization fraction, and the slag film thickness results of the N0 and N1 slags, for the static low-temperature samples, the qt2 and t2 of the N0 slag were smaller than that of the N1 slag own to the low crystallization activation energy of N0 slag. The crystallization fraction and slag film thickness of the N0 slag were larger than that of the N1 slag. For the stirring low-temperature samples, the qt2 and crystallization fraction of the N0 slag and N1 slag were almost the same. The t2 and slag film thickness of the N1 slag were lower than that of the N0 slag. In N1 slag, the shear stress could not only increased the ion immigration speed, but also broken the weak structure units, which further improved the crystallization kinetics conditions. Eventually, a faster crystallization speed was achieved. Additionally, the melting point of N1 slag was smaller than that of N0 slag, which led to the decrease of the slag film thickness of N1 slag.

Table 5. Crystalline fraction, qt2, slag film thickness, and t2 of each slag films with and without agitation at 1198°C.
SampleCrystalline fractionq at t2 (MW/m2)Slag film thickness (mm)t2(s)
Wide sideNarrow side
N0Static low-temperature sample86%0.6053.23.422
Stirring low-temperature sample92%0.4853.53.818
N1Static low-temperature sample72%0.6872.73.024
Stirring low-temperature sample89%0.4912.93.216

3.2.4. Microstructure of Solid Slag Film

The crystal morphology of each low-temperature samples are shown in Fig. 7. As shown in Figs. 7(a) and 7(c), the average grain size of the static low-temperature samples of N0 and N1 slags were 46 um and 55 um respectively. In Figs. 7(b) and 7(d), the average grain size of the stirring low-temperature samples of N0 and N1 slags were 19 um and 18 um respectively, which were almost the same. For the static samples, the crystal density of the samples of N0 slag was larger than that of the samples of N1 slag. For the stirring samples, the crystal density of the samples of N0 slag was similar with that of the samples obtaining from N1 slag.

Fig. 7.

Microstructure of low-temperature samples of the N0 and N1 slags under static and stirring conditions (a) Static low-temperature sample of N0, (b) Stirring low-temperature sample of N0, (c) Static low-temperature sample of N1, (d) Stirring low-temperature sample of N1.

3.3. Air Gap Layer

Many studies1,20,22) pointed out that heat transfer performance from steel shells to mold was significantly affected by the interfacial thermal resistance (air gap layer). The air gap layer was formed by the volume shrinkage of slag films when liquid slag films converted into solid slag films and the glass converted into crystals. It could be represented by the surface roughness (Ra) of slag films.35) Figure 8 shows the typical surface morphology and three-dimensional images of slag films obtained by CLSM. The Ra was obtained by taking the average value of ten data points. The data was extracted by manually “drawing lines” (as shown in Fig. 8) using the professional software (LMeye) equipped for CLSM. In addition, the shedding time of slag films obtained in above heat flux experiment could also characterize the Ra of samples to some extent, which the longer of the shedding time meant the larger of the Ra. Table 6 lists the results of the shedding time and Ra of each samples in this work. It can be found that the shedding time and Ra of the samples with agitation was significantly greater than that of the samples without agitation for both N0 and N1 slags. The more content of crystal were existed in slag films under stirring condition, which led to an increase of the shedding time and Ra. For the static samples, the shedding time and Ra of the samples of N0 slag was larger than that of the samples of N1 slag. For the stirring samples, both the shedding time and Ra of the samples for each slags were nearly the same.

Fig. 8.

Typical surface morphology of slag films obtained by CLSM. The insets represent the corresponding magnified outlines and three-dimensional confocal images. (Online version in color.)

Table 6. The shedding time and Ra of each slag films with and without agitation at 1198°C.
SamplesShedding time (s)Ra (um)
N0Static low-temperature sample4838.51
Stirring low-temperature sample6154.49
N1Static low-temperature sample4131.46
Stirring low-temperature sample5952.87

3.4. Lubrication Property

Break temperature is the temperature at which an extremely increase of the viscosity of mold fluxes was occurred. It was usually used to characterize the lubrication performance of mold fluxes between the steel shell and mold.23) Figure 9 shows the break temperature of the N0 and N1 slags. It could be found that the break temperature of the N1 slag was 9 K lower than that of the N0 slag, which was caused by adopting the low-basicity strategy.

Fig. 9.

The break temperature of N0 and N1 slags. (Online version in color.)

4. Conclusions

A new non-Newtonian mold flux for peritectic steels casting was prepared. The heat transfer behavior and lubrication property of this non-Newtonian mold flux were evaluated and compared with those of the conventional mold flux for peritectic steels casting. The main conclusions were summarized as follows:

(1) For the liquid layer, the structure of N1 slag was broken by shear stresses, which led to a decrease of the lattice heat transfer and radiative heat transfer. The qmax and DOP of N1 slag under stirring condition were reduced to 0.716 MW/m2 and 0.583, respectively. However, the structure of N0 slag had no effect with the shear stress.

(2) For the solid slag layer, the crystallization rate, crystallization fraction, and slag film thickness of each samples were improved by shear stress, which eventually led to a decrease of qt2 (N0 slag: 0.605 MW/m2 to 0.485 MW/m2 and N1 slag: 0.687 MW/m2 to 0.491 MW/m2). However, under stirring condition, the slag film thickness of N1 slag (Wide: 2.9 mm, Narrow: 3.2 mm) was lower than that of N0 slag (Wide: 3.5 mm, Narrow: 3.8 mm). The magnitude of the increase of crystallization rate of N1 slag by shear stress was greater than that of N0 slag. Moreover, the grain size was decreased by the agitation process.

(3) For the air gap layer, Ra and the shedding time of each samples were boosted by shear stress. Ra and the shedding time of N0 and N1 slags under stirring condition were increased to 54.49 um, 61 s and 52.87 um, 60 s, respectively. Therefore, the “low-basicity and high heat-control mold flux” for the peritectic steels casting was came true.

(4) The break temperature of N1 slag (1485 K) was 9 K lower than that of the N0 slag (1494 K). Therefore, the lubrication performance of N1 slag was better than that of N0 slag.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (No. 51574050).

References
 
© 2020 by The Iron and Steel Institute of Japan
feedback
Top