ISIJ International
Online ISSN : 1347-5460
Print ISSN : 0915-1559
ISSN-L : 0915-1559
Regular Article
Development of New Mold Flux for Continuous Casting Based on Non-Newtonian Fluid Properties
Keiji WatanabeKoichi TsutsumiMakoto SuzukiHiroki FujitaSatoshi HatoriTakayuki SuzukiTomoaki Omoto
Author information

2014 Volume 54 Issue 4 Pages 865-871


Due to the importance of the physical and chemical properties of the mold flux used in the production of high-quality steel, in particular the suppression of surface defects on steel sheets, steelmaking engineers have attempted to develop new types of mold flux. This paper presents the results of research on entrapment of mold flux and on heat transfer between the mold and the solidified shell. The authors have been developing a mold flux with non-Newtonian fluid properties using nitrogen. That is, the viscosity of the molten mold flux is low at a high shear rate to reduce the friction between the mold and the solidified shell, but is high at a low shear rate to prevent mold flux entrapment. In order to approximate the properties of mold flux as a non-Newtonian fluid, nitride is added to the conventional flux to adjust the silicate network structure through the reaction between nitrogen and calcium. The contact angle of the non-Newtonian mold flux, which represents the wettability between the mold and the solidified shell, is low in comparison with that of the normal mold flux without nitrogen. It is suggested that the non-Newtonian mold flux increases the heat transfer between the mold and the solidified shell. A casting test was carried out using this non-Newtonian mold flux, and the results showed that entrapment of mold flux decreased and heat transfer increased, as assumed.

1. Introduction

Mold flux plays an important role in the continuous casting process for high-quality steel production. It is well known that mold flux entrapment causes surface defects in steel sheets, and high viscosity mold flux prevents its entrapment in the molten steel. Many studies have been carried out with the aim of preventing mold flux entrapment in the continuous casting mold. One approach is to control the molten steel flow in the mold by electromagnetic force,1,2,3,4) for which several technologies have been proposed, including the use of a static electromagnetic field,1,2) traveling magnetic field3) and stirring magnetic field.4) These electromagnetic methods of controlling molten steel flow have been applied at many steel plants. Another approach is to optimize the shape of the submerged entry nozzle5,6,7,8) to prevent mold flux entrapment. Modifications to the shape at the outlet and in the interior of the submerged entry nozzle have been applied. In another approach, which does not involve control of the molten steel flow, the properties of the mold flux have been newly designed. It is generally considered important to increase the viscosity and interfacial tension between the molten steel and mold flux to prevent entrapment of the mold flux.9,10) The viscosity control technique is used in particular.

At the same time lubricating properties are also required in order to avoid the operational “break-out” problem, which is caused by insufficient lubricant between the mold and the solidified shell. Thus, a low viscosity flux is desired from the viewpoint of stable operation of the continuous caster, while, high viscosity is desired to suppress entrapment of the flux. Therefore, this paper describes the development of a mold flux with non-Newtonian fluid properties using nitride.11) Figure 1 shows the concept of such a mold flux in the continuous casting process. Mold flux is easily entrapped when the surface flow velocity of the molten steel increases. It is highly possible that this entrapment phenomenon occurs near the center between the immersion nozzle and the short side of the mold, because the surface velocity is high at this area.12) On the other hand, a lubrication area is a crevice part in which flux flows in between the mold and the solidified shell. Since the two phenomena occur in different parts of the mold, it is expected to be possible to satisfy both the properties of “lubrication” and “hard to entrap” by controlling the viscosity of the flux at these respective locations.

Fig. 1.

Concept of mold flux of non-Newtonian fluid in a continuous casting process.

A non-Newtonian fluid property means that the viscosity of the molten mold flux is low at a high shear rate, which is advantageous for reducing the friction between the mold and the solidified shell, thereby satisfying “lubrication,” but is high at a low shear rate, which prevents mold flux entrapment and satisfies the “hard-to-entrap” property.

2. Concept of Non-Newtonian Fluid Mold Flux

According to the theory of hydrodynamics, a Newtonian fluid is defined as:   

τ=μ dv dz (1)

In this equation, τ, μ, v and z denote shear stress, viscosity, velocity and displacement, respectively. When this relationship does not hold, the fluid has non-Newtonian properties.

Dilatant flow, Bingham flow, etc. are typical non-Newtonian behaviors. In this study, a pseudoplastic flow, which decreases the viscosity in the high shear rate region, is desired.

Research on the non-Newtonian flow of liquid slag was carried out by Shiraishi,13) who proposed two mechanisms.14,15) One is the dispersion of high melting point particles, for instance Cr2O3, in the slag, which then behaves as a suspension, resulting in non-Newtonian properties. The other is changing the silicate network structure to a hetero network. In the latter case, nitrogen exists with a hetero structure such as the bonding of SiN=Si= or Si N | Si rather than existing in a uniform network as SiO42–. This causes the slag to behave as a non-Newtonian fluid.

3. Experimental Method and Conditions

3.1 Preparation of Mold Flux Containing Nitrogen

In general, various methods of charging nitrogen into slag are possible, for example, by blowing nitrogen gas, by adding Si3N4 powder, and so on. In the present study, a method of adding Si3N4 powder to the mold flux was employed. Figure 2 shows the flow of the procedure for preparing the non-Newtonian flux.

Fig. 2.

Method of preparing the non-Newtonian mold flux.

The melting process is required in order to dissolve nitride in slag. In actual processes, mold flux is used in a powder or granule form, so a water granulation and pulverization process is suitable from the viewpoint of industrial application. It is known that blast furnace slag is foamed by generation of N2 and H2 gases when the slag comes into contact with steam in the water granulation process.16) Because the composition of this mold flux is similar to that of blast furnace slag, which mainly contains CaO and SiO2, it is important to confirm the yield of nitrogen when considering an industrial process.

Therefore, 3.8 kg of silicon nitride was added to 252.5 kg of the main material of the mold flux (CaO: 48%, SiO2: 43%, Al2O3: 3%). The mixture was heated, melted and water granulated under the conditions shown in Table 1, and the yield of nitrogen was estimated by the amount of nitrogen in the flux product. The Kjeldahl method was adopted for the analysis of nitrogen.17) Since there was no large reduction in yield in the granulation processing, as described later, the investigation of properties and the casting test were carried out after the flux was prepared by the procedure shown in Fig. 2.

Table 1. Test conditions for melting Si3N4 powder.
The amount of melting (kg)256
Melting time(min)30–138
Water/flux ratio20–40

Table 2 shows the chemical composition of a conventional industrial mold flux and the non-Newtonian fluid mold flux used in the present work. The conventional mold flux is used for high-speed casting of ultra-low carbon steel.

Table 2. Chemical composition of mold fluxes.

In the next step, the viscosity and contact angles of the fluxes were measured as essential physical properties of non-Newtonian fluid mold flux. Viscosity was measured with the rotational viscometer shown in Fig. 3. A graphite crucible with a radius of 21 mm was filled with 200 g of mold flux. The specimen was heated in an electric furnace to 1573 K under a purified Ar flow. Viscosity was measured by changing the rotational speed (5–300 rpm). Due to concerns about a rise in torque in the low shear velocity region, before measuring the viscosity of the mold flux, the corrected value of viscosity was measured using commercial standard silicon oils (JS-100, 200, 500, manufactured by Nippon Grease Co., Ltd.).

Fig. 3.

High-temperature viscosity meter using concentric cylinder method.

Moreover, the representative values of the shear velocity of the entrapment area in this study were calculated by the flow velocity of molten steel and the molten pool thickness, whose values are assumed to be 20 cm/s12) and 10 mm,18) respectively.   

dv dz = 0.2m/s 10mm =20   1/s
Therefore, in this research, a shear rate of 20 1/s at the entrapment area was used as a representative value.

The shear rate of the lubrication area is calculated by the following equation from the thickness of the film (0.75 mm)19) and the casting conditions (casting speed Vc: 1.6 m/min = 26.7 mm/s, frequency f: 160 cpm = 2.7 Hz, and amplitude A: ±4 mm; oscillating system: sin).   

| V m + V c | d f = | 21.3πcos(2πft)+26.7 | 0.75

Vm: mold speed (mm/s)

Vc: casting speed (mm/s)

f: frequency (Hz)

df: thickness of film (mm)

Thus, the shear velocity changes between 0–125 1/s in one oscillation period.

Various lubrication mechanisms at the boundary of the mold and the strand have been proposed.20,21) However, it is considered that flux flows in the positive strip period of oscillation, and during this period, the shear rate is maximum; therefore, in this research, a shear rate of 125 1/s at the lubrication area was used as a representative value.

The contact angle between a pure iron sheet and the mold flux was also measured using the sessile drop technique to study wettability as an important factor for lubrication. As it is known that oxygen in steel strongly affects interfacial tension,10,22) the oxygen content of the pure iron sheet (10 mm2) was under 10 ppm. The specimen was rapidly cooled by pouring molten mold flux on a water-cooled copper plate and forming 5-mm cubes by polishing the solidified mold flux. These cubes were set in a high-temperature microscope and heated to the target temperatures (1373, 1473, 1573, 1673 K) at a rate of 5 K/min, and were kept at the target temperature for 10 min, then photographed through the quartz window of the high-temperature microscope. From the profiles of mold flux droplets, the contact angle of each flux against the steel sheet was measured.

3.2. Laboratory Continuous Caster Test

Molten steel (250 kg) was cast using a small laboratory continuous caster (mold size: 100 mm in diameter and 310 mm in length) with conventional mold flux and the non-Newtonian mold flux. The cast billet length was about 0.8 m, and the carbon content of the molten steel was selected to be about 0.2%. The oscillation stroke was 8 mm, and the casting velocity was from 1 to 2 m/min. After the casting experiment, the mold flux film was collected from the surface of the mold and cast billet. Flux consumption was calculated from the weight and thickness of the mold flux film.

4. Experimental Results and Discussion

4.1. Nitrogen Yield of Mold Flux

Figure 4 shows the results for the nitrogen yield. Although the nitrogen content decreased gradually with processing time, the content was maintained at about 80% of the initial value until 138 min. Part of the dissolved nitrogen generated fine gas bubbles after heat treatment at 1873 K and water granulation, and as a result, the bulk density of the flux grains was reduced by 40% or more compared with non-N additives. However, no large decrease in yield was observed.

Fig. 4.

Relationship between melting time and nitrogen content.

According to Fuwa et al., the nitrogen content of slag that contains about 800–1000 ppm of nitrogen is reduced by approximately one-half during water granulation.16) Nitrogen evolves as shown in Eq. (2)23) when the slag is placed in contact with steam. However, because the water/slag ratio in this research was 20–40 in mass, which is large in comparison with water granulation of blast furnace slag, the cooling rate was fast enough to avoid nitrogen evolution. Thus, the contact time between the water and slag was short, and as a result, nitrogen yield was high.   

2 N 3- + H 2 O=3 H 2 + N 2 +3 O 2- (2)

4.2. Measurement of Viscosity by Rotational Viscometer

Figure 5 shows the results of viscosity measurement using the concentric cylinder method. The measured data for the viscosity of the conventional mold flux was constant (0.48 Pa·s), while the data for the non-Newtonian mold flux changed from 0.5 to 0.7 Pa·s in proportion to the shear rate. In particular, the low shear rate (20 1/s) was assumed to be the area of entrapment at the meniscus, and the high shear rate (125 1/s) was assumed to be the area of lubrication between the mold and the solidified shell. The viscosity of the non-Newtonian mold flux at the low shear rate was higher than that of the conventional mold flux, and its viscosity at the high shear rate was the same as that of the conventional mold flux. Therefore, there should be no possibility of break-out between the mold and the solidified shell, and the entrapment of mold flux in the molten steel is reduced because the viscosity of the flux is high at the low shear rate.

Fig. 5.

Shear rate dependency of viscosity of mold fluxes at 1573 K.

4.3. Observation of Flowing Flux with Small Laboratory Continuous Caster and Measurement of Flux Consumption

Flux consumption was measured with a small laboratory continuous caster using another non-Newtonian fluid mold flux for which the viscosity in the lubricant region was 0.35 Pa·s. Figure 6 shows the flux consumption by the laboratory caster.

Fig. 6.

Results for flux consumption by laboratory caster.

At a 1.6 m/min casting velocity, the observed flux consumption of the non-Newtonian fluid mold flux was 0.50 kg/m2 and that of the conventional mold flux was 0.38 kg/m2. That is, the flux consumption of the non-Newtonian fluid mold flux increased by more than 20% in comparison with that of the conventional mold flux. Figure 7 shows a view of the lubricant as photographed from under the mold. Based on observation through the video camera under the mold of the laboratory continuous caster, the wettability of the non-Newtonian fluid mold flux increased.

Fig. 7.

Observation of lubricant (photograph taken under mold of laboratory continuous caster).

Figure 8 shows the measurement results for the contact angle between the pure iron sheet and the mold flux in the high-temperature microscope. The contact angle of the non-Newtonian fluid mold flux was smaller than that of the conventional mold flux at high temperature. These results showed that the wettability between the molten mold flux and the pure iron sheet was improved with the non-Newtonian fluid mold flux.

Fig. 8.

Temperature dependency of contact angle of mold fluxes.

In explaining this phenomenon, since the flux contained nitride, it is presumed that part of the nitrogen in the flux was absorbed to the steel surface, and this interaction influenced the wettability.24) In this manner, consumption of the non-Newtonian mold flux in the small casting test increased by about 20% in comparison with the conventional flux with a similar viscosity in the lubrication region. This fact indicated the excellent wettability of the non-Newtonian fluid flux.

5. Actual Casting Test

An actual casting test was carried out in order to confirm the effect of the non-Newtonian fluid mold flux using two strands. In one strand, the trial mold flux (hereafter, “conventional flux”) was used until casting was completed. In the other strand, the conventional flux was changed to non-Newtonian fluid flux at the casting length position of 700 m.

The material being cast was steel for oil well tubulars. The strand size and casting speed were 210 mmφ and 1.4–1.6 m/min, respectively.

Flux consumption was evaluated as an index of lubrication. To measure entrapment of mold flux, four specimens with dimensions of 20 mm × 40 mm × 40 mm were cut and taken from the strand at the casting length positions of 700 m and 800 m. In the test strand, the 700 m position was before the flux change, i.e., while using the conventional flux, and the 800 m position was after changing to the non-Newtonian fluid flux.

Flux inclusions were investigated by sampling from the strands at each length position (Fig. 9).25) The viscosity of the non-Newtonian fluid flux in the low shear zone shows a difference of 0.1 Pa·s or more, even though the viscosity of both the conventional flux and the non-Newtonian fluid flux in the lubrication region is at the same level, 0.81–0.82 Pa·s.

Fig. 9.

Schematic of flux analysis position at the sampling strand.

Figure 10 shows the consumption of each flux: the conventional flux was 0.13 kg/m2 and the non-Newtonian fluid flux was 0.14 kg/m2. Although no clear difference like that in the laboratory casting test was observed in the actual casting test, the tendency of flux use was almost the same as in the laboratory caster test.

Fig. 10.

Flux consumption (Vc = 1.6 m/min) in test casting.

Figure 11 shows the entrapment amount for each flux. The conventional flux was used at “strand A” and the non-Newtonian flux was used at “strand B (after 700 m)”, respectively. When elements (Si, Ca, Al, and Na) that are constituents of the flux were detected in the residue after galvanostatic electrolysis, the residue was assumed to be a flux inclusion.

Fig. 11.

Mold flux entrapment in test casting.

At the 700 m position, that is, where both strands were using the conventional flux, the entrapment in strand B was 13% lower than in strand A. This difference is presumably due to the different characteristics of strand A and B, such as the difference of inclusion adhesion on the respective immersion nozzles. However, at the 800 m position, where strand B was using the non-Newtonian flux, the entrapment in strand B decreased to less than one-half of that in strand A. Because the reduction of entrapment in strand B (>50%) is substantially larger than the initial difference between the strands (13%), the non-Newtonian flux was considered to display a “hard-to-entrap” property.

Figure 12 shows the heat flux in the mold calculated from the circulating water temperature and the flow velocity. The heat flux of the non-Newtonian type was 1.2 times that of the conventional flux.

Fig. 12.

Heat transfer in the billet mold.

A phenomenon in which mold flux adheres to the cast strand, as observed in the laboratory caster, causes the heat flux to increase. It is well known that the interfacial thermal resistance between a flux film and a mold has an influence on the heat flux.26) In the actual casting test using the non-Newtonian fluid flux, it is considered that the wettability with steel improved, as in the laboratory caster test, enabling uniform flux flow between the mold and the strand. This caused the air gap to decrease, and as a result, the heat flux increased.

6. Mechanism of Non-Newtonian Fluid Mold Flux

Finally, the mechanism of the non-Newtonian fluid mold flux was considered based on the results of these laboratory experiments. First, the possibility of the effect of a suspension of high melting point particles, which was the first proposed mechanism mentioned in Chapter 2,15) was examined by an SEM investigation of the flux film sampled in the laboratory caster, as described in section 4.3. Figure 13 shows a photograph of the flux film, which has a glassy structure, as in the case of the conventional flux film, and also contains no Si3N4 particles. Figure 14 shows its X-ray-diffraction pattern. As shown in the XRD chart, no peak of the crystal of Si3N4 was observed. These facts show that nitride was melted into the flux. From this, it was presumed that the non-Newton property observed in this study was influenced not by suspension of Si3N4 particles, but rather, by dissolved nitrogen.

Fig. 13.

Photograph of mold flux film.

Fig. 14.

XRD results of the flux film in laboratory caster.

Next, the mechanism of the change in flux viscosity with shear rate when the mold flux contained dissolved nitride was assumed to be as follows.13,15) The viscosity of the non-Newtonian mold flux remains high because there are weak bonds between the nitrogen and calcium in the molten non-Newtonian fluid mold flux at the low shear rate. However, at the high shear rate, these weak bonds between the nitrogen and calcium in the molten non-Newtonian fluid mold flux are broken, and as a result, its viscosity decreases.

7. Conclusion

A non-Newtonian fluid mold flux containing nitrogen was developed for the continuous casting process of high-quality steel. The basic properties of this flux such as viscosity and wettability were measured, and casting tests were performed in the laboratory and with an actual continuous caster. The following conclusions were obtained.

(1) Viscosity of the mold flux containing nitrogen is high at a low shear rate, and low at a high shear rate.

(2) Wettability between the molten mold flux and steel is improved with the non-Newtonian mold flux, resulting in improved lubrication.

(3) The actual casting test using the non-Newtonian flux clarified the fact that inclusions of the non-Newtonian flux in the strand decrease, even though the lubricant properties of the two fluxes are on the same level. This difference in inclusions is attributed to the higher viscosity of the non-Newtonian flux under the low shear rate condition in the entrapment region.

Therefore, this flux enables the production of high-quality slabs by preventing entrapment of mold flux during high-speed casting, while its high lubrication property in the high shear rate region reduces the frequency of the operational problem known as break-out, which is caused by inadequate lubrication.

© 2014 by The Iron and Steel Institute of Japan