Improving Thermal Insulation Of Magnesia Dry Mix & Reducing Refractory Consumption Per Ton Steel

At present, magnesia dry vibratable mixes are the primary refractory materials used for the working linings of continuous casting tundishes. These dry mixes exhibit excellent resistance to corrosion from high-iron and basic slags, long service life, no contamination to molten steel, convenient construction, and easy stripping and turnover. They have gradually been applied in tundish working linings with outstanding performance and promising application prospects.

However, magnesia dry mixes generally have a high bulk density, leading to the following disadvantages:

High thermal conductivity: The heat dissipation rate of the tundish lining is fast, resulting in a rapid temperature drop of molten steel and high energy consumption.

High refractory consumption: The high density leads to greater refractory consumption per ton of steel, which wastes mineral resources and increases refractory costs for enterprises.

Thermal expansion and poor resistance: Magnesium oxide has a large thermal expansion coefficient, poor thermal shock resistance, and is prone to moisture absorption and hydration. Additionally, it has poor resistance to thermal spalling and structural spalling, which significantly reduces the service life of the refractory material.

Furthermore, current tundish dry mixes are primarily bonded with phenolic resin. During the baking process, as the tundish temperature rises, the phenolic resin gradually cures to form a carbon network structure, providing the dry mix with sufficient strength. However, phenolic resin-bonded dry mixes result in carbon pickup in molten steel, which significantly affects the quality of clean steel production. Moreover, the cured resin decomposes between 200°C and 800°C, releasing gases such as CO₂, CO, CH₄, H₂, and H₂O. The free phenol and generated gases form irritating flue gas, which is harmful to the health of on-site workers.

Therefore, the existing refractories for tundish working linings no longer meet the requirements for cleanliness and thermal insulation performance.

This paper focuses on the research and development of a resource-saving, lightweight, and eco-friendly tundish dry mix with low apparent bulk density. It not only reduces refractory consumption per ton of steel and improves thermal insulation properties but also uses an environmentally friendly binder that produces no harmful or irritating gases, ensuring strong environmental performance.

01. Experiment

1.1 Raw Materials and Experimental Scheme

The main raw materials used in this experiment and their chemical compositions are listed in Table 1.

Sintered magnesia, forsterite, and magnesite were used as the primary raw materials (their physical and chemical properties are shown in Table 1). The particle sizes of 5–1 mm, ≤1 mm, and 200 mesh were blended in a specific ratio, with binder, additives, and a small amount of paper fiber added. The experimental formulation is given in Table 2.

1.2 Test Procedure and Properties Measurement

According to the designed formulation, a special iron cup was used as the container, and its mass m1 was weighed. Each group of well-mixed materials was placed on the bulk density tester. The iron cup was positioned at the bottom, filled with materials, and the surface was smoothed to achieve natural loose packing. The mass m2 was then weighed. The sample mass was calculated as m=m2−m1, and the bulk density was determined by:

Bulk Density= Mass m/Volume V

Each well-mixed material was further blended in a mixer, then manually rammed into 160 mm × 40 mm × 40 mm triple cement specimen molds. The cuboid specimens were formed by spreading and compacting layer by layer until full. The rammed specimens, together with the molds, were dried in an oven at 200°C for 3 hours, then demolded after cooling.

The bulk density, cold crushing strength, and linear change on firing of the specimens after heat treatment at 200°C for 3 hours and 1500°C for 3 hours were measured to evaluate the low-temperature bonding behavior and sintering properties. In addition, the prepared dry mix was manually rammed into disc molds with a diameter of 180 mm and thickness of 20 mm, cured at 200°C for 3 hours, and demolded. The thermal conductivity at different temperatures was then tested.

02 Results and Discussion

2.1 Bulk Density and Apparent Porosity Density

Figure 1 shows the bulk density and apparent porosity density of the dry mix.

It can be seen from Figure 1 that after replacing phenolic resin powder with glucose, the bulk density of the dry mix changes slightly, while the apparent porosity density after heating at 200°C for 3 hours decreases slightly. When 6 wt% glucose is added, the apparent porosity density is reduced by approximately 5%. However, with the addition of magnesite and forsterite, both the bulk density and the apparent porosity density after heating at 200°C for 3 hours show little change. With the introduction of paper fiber, both the bulk density and apparent porosity density after heating at 200°C for 3 hours decrease significantly. When 0.1 wt%, 0.2 wt%, and 0.3 wt% paper fiber are added, the bulk density is reduced by 2.7%, 5.4%, and 8.1%, respectively, and the apparent porosity density is reduced by 1.3%, 2.6%, and 5.7%, respectively.

Comparing Sample G1 and Sample G12, considering glucose as the binder, with 40 wt% forsterite and 0.3 wt% paper fiber added (Sample G12), both the bulk density and apparent porosity density are reduced by approximately 9%. The reduction in bulk density and apparent porosity density enables lightweighting of the material, lowers refractory consumption per ton of steel for the tundish, and saves refractory costs for enterprises.

It can also be seen from Figure 1 that with the increase in magnesite addition, the apparent porosity density after heating at 1500°C for 3 hours shows a downward trend.

Figure 1 shows that the introduction of magnesite and forsterite, which have relatively low apparent porosity density, has little effect on the bulk density and apparent porosity density of the dry mix after heating at 200°C for 3 hours. It is considered that the apparent porosity density of magnesite and forsterite is about 7.9% lower than that of dead-burned magnesia. Based on the 40 wt% addition ratio, the apparent porosity density could theoretically be reduced by about 3.1%. However, there are many factors affecting apparent porosity density, such as the particle shape of raw materials. Particle shape is related to the structural characteristics and crushing method of the raw materials, so the particle shape and size vary. Therefore, its effect on the apparent porosity density of the dry mix is not obvious in this study and requires further investigation.

Using glucose as a binder has a significant effect on reducing both the bulk density and apparent porosity density of the dry mix. This is mainly due to the difference in particle size: glucose particles are coarser than resin powder, while phenolic resin powder has a finer particle size and can fill pores more uniformly. It can also be seen from the figure that the apparent porosity density after heating at 1500°C for 3 hours shows a trend similar to that after heating at 200°C for 3 hours. However, Samples G-5 to G-8, which have magnesite introduced, exhibit a significant decrease in apparent porosity density after heating at 1500°C for 3 hours, showing a linear decline with increasing magnesite addition ratio. This is because magnesite begins to decompose at about 400°C during the heating process, releasing CO₂ gas. The reaction is most intense between 550°C and 650°C and completes at 1000°C, forming light-burned MgO with a loose texture, high porosity, and low apparent porosity density, thus reducing the apparent porosity density of the dry mix after heating at 1500°C for 3 hours.

2.2 Physical Properties at Room Temperature

Figure 2 shows the effect of different raw materials on the physical properties of the dry mix. It can be seen from the figure that, compared with phenolic resin, the specimens using glucose as the binder exhibit lower cold crushing strength after heat treatment at 200°C for 3 hours and at 1500°C for 3 hours. The cold crushing strength of the specimens increases gradually with an increase in glucose addition. When glucose is added at 6 wt%, although the strength of the bonded specimens is still not comparable to that of specimens bonded with phenolic resin, it can fully meet the demolding requirements on the customer's site.

Phenolic resin forms a relatively strong three-dimensional network structure after low-temperature heat treatment. Glucose has a melting point of 146°C and can undergo cross-linking and condensation reactions during baking at 200°C to form a three-dimensional network structure. In addition, water produced by the decomposition of glucose during heating can also react with magnesia, contributing to the strength development of the dry mix. Glucose is non-toxic and harmless, and does not release irritating gases such as phenol and formaldehyde during baking. It has the advantages of being eco-friendly, green, and low-cost, making it an environmentally friendly binder to replace phenolic resin.

It can also be seen from Figure 2 that when the glucose addition is further increased to 7 wt%, the increase in the cold crushing strength of the dry mix is not significant, while the linear change rate after heat treatment at 1500°C for 3 hours becomes relatively large. Therefore, the recommended addition ratio of glucose is 6 wt%.

With an increase in the magnesite addition ratio, the cold crushing strength of the specimens after heating at 200°C for 3 hours changes slightly, while the crushing strength after heat treatment at 1500°C for 3 hours shows a gradual decreasing trend, and the linear change rate presents the same tendency. This is mainly caused by the gas released from the decomposition of magnesite during baking, which results in a loose texture and high porosity.

With an increase in the paper fiber addition ratio, the crushing strength of the specimens after heat treatment at 200°C for 3 hours and at 1500°C for 3 hours both show a decreasing trend. It is analyzed that the introduction of paper fiber makes it difficult for the material to achieve dense packing, thus affecting the binder's effectiveness. Furthermore, after volatilization during baking, pores form inside the material, leading to the degradation of its physical properties. When the paper fiber addition is 0.3 wt%, the room-temperature physical properties can still meet the on-site demolding requirements. Additionally, as the paper fiber shrinks and burns during baking, micro-pores and channels form inside the dry mix, which facilitate the rapid discharge of gas generated by binder decomposition during heating, relieve internal stress, and reduce the risk of working lining collapse.

It can be seen from Figure 1(b) that with an increase in the magnesite addition ratio, the linear change rate after heat treatment at 1500°C for 3 hours increases gradually. This is attributed to the gas release from magnesite decomposition during baking, the loose texture, and sintering shrinkage during high-temperature heat treatment. For specimens with added forsterite, the linear change rate after heat treatment at 1500°C for 3 hours is significantly reduced. It is speculated that MgO further reacts with SiO₂ to form the forsterite phase, whose micro-expansion can fill pores and reduce the linear change rate. Meanwhile, it can also improve the slag penetration resistance of the material.

2.3 Thermal Conductivity

The thermal conductivity comparison of Samples G-1, G-3, G-9, and G-12 at different temperatures is shown in Figure 3. It can be seen that the thermal conductivity exhibits a decreasing trend at all temperatures. The thermal conductivity of G-3 is significantly lower than that of G-1 at 600°C, 800°C, and 1000°C. This is because phenolic resin forms a carbon network structure after carbonization, which enhances thermal conductivity, whereas G-3 uses glucose as the binder, resulting in much lower residual carbon and, consequently, lower thermal conductivity.

Compared to Sample G-3, the thermal conductivity of Sample G-9 is further reduced. This is attributed to the lower thermal conductivity of forsterite sand compared to magnesia (6.7 W·m⁻¹·K⁻¹ at 1000°C). In Sample G-12, the introduction of paper fiber leads to a decrease in bulk density and the formation of micro-pores during baking, which increases porosity, reduces the formation of low-melting substances, effectively prevents the aggregation of low-melting phases, and makes sintering more difficult, thus reducing thermal conductivity.

A lower thermal conductivity is beneficial in two aspects. On one hand, it helps form a large temperature gradient from the hot face to the cold face of the working lining, enabling gradual sintering from the working lining to the permanent lining and forming a dense structure on the hot face of the working lining. This, in turn, eliminates through-cracks and other defects that may cause slag penetration into the permanent lining. On the other hand, it reduces the heat dissipation rate of the tundish lining, improves thermal insulation performance, lowers the shell temperature, slows the temperature drop of molten steel during use, and saves a significant amount of heat energy.

03 Conclusions

(1) Using non-toxic and harmless glucose instead of phenolic resin as the binder not only provides sufficient demolding strength for the material, but also avoids the release of irritating gases such as phenol and formaldehyde during baking. It is environmentally friendly, low-cost, and green.

(2) The introduction of paper fiber can reduce the bulk density, apparent porosity density, and thermal conductivity of the dry mix, achieving a lightweight design for the material and contributing to energy conservation and reduced consumption.

(3) Using glucose as the binder and introducing forsterite and paper fiber are both beneficial in reducing thermal conductivity, decreasing the heat dissipation rate of the tundish lining, and improving its thermal insulation performance.