In my research, I focused on solving two critical manufacturing problems associated with cast steel cooling walls for blast furnaces: the deformation and melting of embedded cooling water pipes during pouring, and the severe sand foundry defects that appear on the external surface after solidification. These two issues have long hindered the large-scale production and application of cast steel cooling walls. Through numerical simulation, experimental investigation, and production validation, I developed effective anti-penetration techniques and a high-performance anti-sand coating. In this article, I will present my systematic research and the main conclusions drawn from this work.
Blast furnace cooling walls are essential components that protect the furnace shell and refractories from high-temperature damage. Among the various types of cooling walls, cast steel cooling walls offer excellent elongation, high tensile strength, high melting point, and good thermal shock resistance. Compared with cast iron cooling walls, they provide superior thermal conductivity because the steel matrix can metallurgically bond with the steel cooling water pipe, eliminating the air gap that normally reduces heat transfer. Compared with copper cooling walls, they are significantly cheaper and more resistant to high-temperature erosion. However, the manufacturing of cast steel cooling walls is challenging. The pouring temperature is extremely high, typically around 1550 °C. At this temperature, the embedded cooling water pipe may easily deform or melt through. In addition, because the casting is heavy and thick, the molten steel remains liquid for a long time, leading to deep penetration into the sand mold and consequently severe sand foundry defects on the casting surface.
My research aimed to address these problems in a systematic manner. First, I carried out a numerical simulation of the filling and solidification process using the finite element method. This allowed me to predict the temperature field of the cooling water pipe and identify the critical locations where melting would most likely occur. Then, I proposed a method of filling the cooling water pipe with solid cooling materials to reduce the pipe temperature and prevent melt-through. I studied the filling and cleaning behavior of resin sand used as the solid cooling medium. Finally, I developed an alcohol-based anti-sand coating with white corundum as the refractory aggregate, and I optimized its composition through both single-factor experiments and orthogonal tests. Production trials confirmed that the proposed methods are practical and effective.
Throughout this article, I will emphasize how the problem of sand foundry defects can be mitigated by combining proper casting design, protective measures for the cooling pipe, and a specially formulated coating. I believe that the results presented here can contribute to the broader application of cast steel cooling walls in the metallurgical industry.
Numerical Simulation of the Casting Process
The cast steel cooling wall investigated in my study had an external dimension of 2100 mm × 745 mm × 595 mm. The wall thickness varied from 190 mm at the thinnest section to 430 mm at the thickest section, with an average thickness of 288 mm. The cooling water pipe had a U-shape with an outer diameter of 60 mm, a wall thickness of 5 mm, and a total length of 5105 mm. The pipe was made of 20G steel, while the cooling wall matrix was ZG35CrMnSi low alloy steel. The total mass of the casting was approximately 2686.4 kg.
I designed a bottom gating system to ensure smooth filling of the casting cavity. The bottom gating system is known for its ability to reduce turbulence and prevent slag entrapment. The cross-sectional area of the gating system was determined using the well-known orifice equation:
$$S_{\text{res}} = \frac{m}{\mu \tau \rho H_p}$$
where \( m \) is the mass of molten metal, \( \mu \) is the flow coefficient, \( \tau \) is the pouring time, \( \rho \) is the density of the liquid metal, and \( H_p \) is the average hydrostatic pressure head. Substituting the values for my cooling wall, I calculated the minimum cross-sectional area as 1256.63 cm². The gating system was designed as a semi-open type with a cross-sectional area ratio of \( S_{\text{runner}} : S_{\text{inner}} : S_{\text{sprue}} = 2 : 1.5 : 1 \). The actual dimensions were as follows:
| Gating component | Quantity | Dimension |
|---|---|---|
| Sprue | 1 | φ60 mm × 700 mm |
| Runner | 1 | φ70 mm × 1950 mm |
| Inner gate | 6 | φ40 mm × 85 mm |
I then used the ProCAST finite element software to simulate the mold filling and solidification processes. The initial temperature of the molten metal was set to 1550 °C. The cooling water pipe and the sand mold were initially at 25 °C. The pouring speed was 30 kg/s under gravity filling. The heat transfer coefficients were set as follows: 500 W/(m²·K) between the casting and the mold, and 1000 W/(m²·K) between the casting and the cooling water pipe. The simulation results showed that the total filling time was about 100.8 seconds. During filling, the maximum temperature of the cooling water pipe remained below 1100 °C, which was far below the solidus temperature of the pipe material, 1473.2 °C. Therefore, the pipe was safe during mold filling.
However, the solidification stage was much more dangerous. The total solidification time was about 22347.3 seconds. During this period, the local maximum temperature of the cooling water pipe rose to between 1502 °C and 1510 °C. Since the liquidus temperature of the 20G steel pipe was approximately 1503.6 °C, the pipe could easily melt through. To identify the exact location of the highest risk, I selected six cross sections on the pipe near the inner gate side, labeled A through F. The temperature histories of these sections are compared in the following table.
| Section | Position relative to inner gate | Peak temperature (°C) | Risk of melt-through |
|---|---|---|---|
| A | Farthest | ~1450 | Low |
| B | Far | <1500 | Low |
| C | Medium | <1500 | Low |
| D | Medium | <1500 | Moderate |
| E | Nearest to thickest part | 1504.1 | High |
| F | Farthest on other side | ~1456 | Low |
From the simulation, I found that section E was the most vulnerable region. The lower surface of the pipe at this section reached 1504.1 °C, which exceeded the melting point of the pipe material. Therefore, the pipe would melt through at the bottom surface in the area corresponding to the thickest part of the cooling wall. In contrast, section F remained below the solidus temperature, meaning that the pipe might not achieve metallurgical bonding with the matrix at that location.
I also studied the effect of pouring temperature on the interface bonding ratio. I defined the bonding ratio as the fraction of the pipe surface area that achieved metallurgical bonding with the matrix. The results are summarized in the table below.
| Pouring temperature (°C) | Bonding ratio (%) | Phenomenon |
|---|---|---|
| 1520 | Some regions below solidus | Incomplete bonding at section F |
| 1550 | 91.2 | Best compromise |
| 1580 | Severe melt-through | Multiple sections melt |
As the pouring temperature increased, the fraction of the pipe that reached the liquidus temperature increased, which improved the bonding ratio. However, when the temperature exceeded 1550 °C, many sections of the pipe melted through, reducing the effective bonding ratio. Thus, I concluded that 1550 °C was the optimal pouring temperature, but additional protective measures were still necessary to prevent local melt-through at section E.
Solid Cooling Material for Pipe Protection
To prevent the cooling water pipe from melting through during solidification, I proposed filling the interior of the pipe with a solid cooling material. The solid material absorbs heat from the pipe and provides a chilling effect to the surrounding molten metal. I chose resin sand as the solid cooling material because it can be tightly packed inside the pipe, has good heat absorption capacity, and can be removed after casting. The effectiveness of this approach was first verified by simulation. When the pipe was filled with solid cooling material, the maximum temperature at section E dropped to 1492.4 °C, which was below the melting point of the pipe material. The minimum temperature at section F was 1476.2 °C, which remained above the solidus temperature. Therefore, the simulation indicated that the pipe would neither melt through nor lose its ability to bond metallurgically with the matrix.
Encouraged by this result, I carried out experiments to determine the best composition and process parameters for the resin sand. The experiments were conducted using a scaled-down cooling water pipe made of the same 20G steel. The filling efficiency was defined as:
$$K_{\text{fill}} = \frac{m_1 – m_0}{\rho V} \times 100\%$$
where \( m_0 \) is the pipe mass, \( m_1 \) is the total mass after filling, \( \rho \) is the compacted density of the resin sand, and \( V \) is the inner volume of the pipe. I compared three kinds of sand: ceramsite, zircon sand, and chromite sand. Their angularity coefficients were 1.05, 1.50, and 1.35, respectively. The filling results are shown below.
| Sand type | Angularity | Filling efficiency (%) |
|---|---|---|
| Ceramsite | 1.05 | 87.4 |
| Chromite sand | 1.35 | 73.2 |
| Zircon sand | 1.50 | 62.1 |
Ceramsite gave the highest filling efficiency because its spherical shape reduced friction and improved flowability. I therefore selected ceramsite as the primary sand for all subsequent experiments.
Next, I investigated the influence of filling pressure. Experiments were performed at pressures of 0.1, 0.3, 0.5, 0.7, and 0.9 MPa. The results are presented in the table below.
| Filling pressure (MPa) | Filling efficiency (%) |
|---|---|
| 0.1 | ~68 |
| 0.3 | ~87 |
| 0.5 | ~98 |
| 0.7 | ~99 |
| 0.9 | ~94 |
Although 0.7 MPa produced the highest filling efficiency, 0.5 MPa was selected as the optimal pressure because it gave nearly identical results while imposing lower requirements on the equipment. The binder content was then optimized from 0.8% to 1.6% by weight of sand. The optimum binder content was found to be 1.2%, which gave a filling efficiency of 99.1%. Higher binder contents caused the sand to agglomerate, reducing its flowability.
After the resin sand was compacted inside the pipe, the pipe was heated to 1495 °C for 35 minutes to simulate the casting environment. After cooling, I attempted to remove the sand by vibration and impact. Only about 60% of the sand could be removed because the sand particles had partially sintered together. To improve the cleaning efficiency, I added fused alumina sand to the original sand mixture. The cleaning efficiency was defined as:
$$K_{\text{clean}} = \frac{m_4}{m_3 – m_0} \times 100\%$$
where \( m_3 \) is the total mass before cleaning, and \( m_4 \) is the mass of sand removed. The results for different alumina additions are given below.
| Alumina content in sand mixture (%) | Cleaning efficiency (%) |
|---|---|
| 0 | ~60 |
| 5 | ~74 |
| 10 | ~88 |
| 15 | 96.8 |
| 20 | ~97.5 |
| 25 | ~98 |
Adding 15% fused alumina by weight of the original sand was sufficient to reach 96.8% cleaning efficiency. Beyond this point, the improvement was marginal. Therefore, I chose 15% alumina as the optimal addition because it balanced cleaning performance with cost.
In the actual production trial, I filled the cooling water pipe with resin sand containing 85% ceramsite and 15% fused alumina. The pipe was fixed in the mold using cylindrical steel supports and spacers. The pouring temperature was controlled at 1550 °C. After pouring and solidification, the cast steel cooling wall was sectioned along the pipe axis. The macroscopic inspection showed that the cooling water pipe was well bonded to the matrix, with no visible gaps, no deformation, and no melt-through. The pipe and the matrix appeared to have fused together homogeneously.
I then examined the metallurgical bonding zone using optical microscopy and scanning electron microscopy. The microstructure of the bonding zone was composed of pearlite and ferrite, similar to the matrix and the pipe, but with slightly different morphology. The ferrite exhibited a needle-like shape due to the rapid cooling effect of the pipe on the molten steel. Elemental line scanning across the interface revealed that silicon, chromium, and manganese had diffused from the matrix into the pipe surface, confirming that a true metallurgical bond had formed. Shear tests were performed on specimens cut from the bonding zone. The average shear strength measured was 318 MPa. The fracture surface showed ductile dimples, indicating that the bond had high strength and toughness. This result demonstrated that the solid cooling material method could achieve both melt-through prevention and a strong metallurgical bond.
Development of Anti-Sand Sticking Coating
The second major problem I addressed was the formation of sand foundry defects on the thick cast steel cooling wall. Because the casting has many thick sections and some grooves, the molten steel remains liquid for a long time during solidification. This increases the tendency for metal to penetrate into the sand mold, resulting in mechanical and chemical sand sticking. I chose to develop a sinterable, easily peelable alcohol-based coating that could be applied to the sand mold surface. The coating was designed to form a dense sintered layer at high temperature, blocking the pores between sand grains and preventing metal penetration.
The coating formulation consisted of white corundum as the refractory aggregate, potassium feldspar powder and iron oxide powder as fluxes, lithium bentonite and polyvinyl butyral (PVB) as suspending agents, phenolic resin and hexamethylenetetramine as binders, Tween-80 as a surfactant, and ethanol as the carrier liquid. The performance of the coating was evaluated through several parameters, including conditional viscosity, density, suspension rate, and thermal shock cracking resistance.
In my first set of experiments, I studied the effect of lithium bentonite content on the coating properties. The suspension rate was calculated using the following formula:
$$\Phi = \left(1 – \frac{v}{100}\right) \times 100\%$$
where \( v \) is the volume of clear liquid layer at the top of a 100 mL graduated cylinder. The results for different lithium bentonite additions are summarized below.
| Lithium bentonite (g) | Conditional viscosity (s) | Density (g/cm³) | 2h suspension (%) | 24h suspension (%) |
|---|---|---|---|---|
| 4 | 15.96 | 1.50 | 84.0 | 73.5 |
| 8 | 16.26 | 1.51 | 89.0 | 74.0 |
| 12 | 16.66 | 1.49 | 96.0 | 79.0 |
| 16 | 20.25 | 1.56 | 97.0 | 83.0 |
As the lithium bentonite content increased, the viscosity increased and the suspension rate improved. However, an excessively high viscosity made the coating difficult to apply. I selected 12 g as the optimal amount based on the balance between viscosity and suspension.
Next, I studied the effect of phenolic resin content. The results are shown in the following table.
| Phenolic resin (g) | Conditional viscosity (s) | Density (g/cm³) | 2h suspension (%) | 24h suspension (%) |
|---|---|---|---|---|
| 2.0 | 15.20 | 1.47 | 92.0 | 79.0 |
| 2.5 | 16.66 | 1.49 | 96.0 | 83.5 |
| 3.0 | 22.22 | 1.48 | 93.5 | 79.0 |
| 3.5 | 21.09 | 1.49 | 93.0 | 78.5 |
The optimal phenolic resin content was 2.5 g, which gave the best combination of viscosity and suspension stability.
After these single-factor experiments, I conducted an L16 orthogonal test to optimize the combined effects of lithium bentonite, phenolic resin, PVB, and ethanol. The four factors and their levels are listed in the table below. In the test formulations, the amounts of white corundum, potassium feldspar, iron oxide, and Tween-80 were fixed at 100 g, 24 g, 12 g, and 0.1 g, respectively. The total powder mass was 136 g.
| Factor | Level 1 | Level 2 | Level 3 | Level 4 |
|---|---|---|---|---|
| Lithium bentonite (wt.%) | 8.3 | 8.8 | 9.3 | 9.8 |
| Phenolic resin (wt.%) | 1.4 | 1.8 | 2.2 | 2.6 |
| PVB (wt.%) | 0.12 | 0.22 | 0.32 | 0.42 |
| Ethanol (wt.%) | 70 | 66 | 62 | 58 |
The conditional viscosity and suspension rates of the 16 test coatings were measured. The range analysis results showed that ethanol had the greatest influence on both viscosity and 24-hour suspension rate. Lithium bentonite also had a significant effect on the 24-hour suspension rate. Based on the orthogonal analysis, the optimal formulation was selected as follows:
| Component | Optimal amount | Percentage of total powder |
|---|---|---|
| White corundum | 100 g | 73.5% |
| Potassium feldspar | 24 g | 17.6% |
| Iron oxide | 12 g | 8.8% |
| Lithium bentonite | 9.8% | 9.8% |
| Phenolic resin | 1.8% | 1.8% |
| Hexamethylenetetramine | 0.36% | 0.36% |
| PVB | 0.12% | 0.12% |
| Ethanol | 66% | 66% |
I then investigated the effect of the flux ratio, i.e., the mass ratio of potassium feldspar to iron oxide, on the thermal shock cracking resistance of the coating. The total flux amount was fixed at 36 g per 100 g of white corundum. The coating applied on cylindrical sand specimens was placed in a furnace at 1300 °C for 2.5 minutes. The cracking resistance was graded from I to IV, where I represented no cracks and IV represented wide cracks. The results are shown below.
| Potassium feldspar : iron oxide | Cracking resistance grade | Surface appearance |
|---|---|---|
| 0 : 1 | III | Branch-like cracks |
| 1 : 3 | II | Fine network cracks |
| 1 : 2 | II | Fine network cracks |
| 1 : 1 | II | Fine network cracks |
| 2 : 1 | I | Smooth, no cracks |
| 3 : 1 | II | Fine network cracks |
| 1 : 0 | III | Branch-like cracks |
| No flux | IV | Severe peeling |
The flux ratio of 2:1 gave the best thermal shock resistance. The presence of potassium feldspar and iron oxide promoted sintering of the white corundum particles, forming a dense layer that prevented both mechanical metal penetration and chemical reaction between the metal and the sand. This dense layer effectively reduced sand foundry defects and allowed the coating to peel off easily from the casting surface after cooling.
Production Validation and Practical Application
To validate the developed coating under real production conditions, I applied the optimal coating onto the sand mold surface of a cast steel cooling wall. The coating was first sprayed and then brushed to ensure uniform thickness, especially in the grooves and corners. After ignition and drying, the mold was ready for pouring. In the actual production environment, an automatic pouring line was used to control the pouring process consistently. An example of such a line is shown below.
After the cast steel cooling wall had solidified and cooled, I performed sand cleaning by simple impact. Most of the sand layer on the surface fell off automatically. Only a few spot-like adhesions remained in the deep grooves of the cooling wall. The cleaning area reached more than 95% of the total surface area. This level of cleanliness satisfied the production requirements and significantly reduced the need for manual grinding.
The combination of the solid cooling material in the cooling water pipe and the anti-sand coating on the mold surface effectively solved the two major manufacturing challenges. The cooling water pipe remained intact and metallurgically bonded to the matrix, while the external surface of the cooling wall was largely free of sand foundry defects. These achievements are important for the reliable production and broader industrial application of cast steel cooling walls.
Conclusions
In my research, I conducted a comprehensive investigation to improve the manufacturing of cast steel cooling walls. The following conclusions can be drawn from my work:
First, numerical simulation showed that during mold filling, the cooling water pipe temperature remained below 1100 °C, which was safe. However, during solidification, the pipe temperature at the thickest section reached 1502–1510 °C, exceeding the melting point of the pipe material. The optimal pouring temperature was found to be 1550 °C, where the metallurgical bonding ratio between the pipe and the matrix reached 91.2%.
Second, filling the cooling water pipe with resin sand containing ceramsite and 15% fused alumina effectively reduced the peak temperature of the pipe to 1492.4 °C, preventing melt-through while still allowing metallurgical bonding. The filling efficiency reached 99.1% when the binder content was 1.2% and the filling pressure was 0.5 MPa. The cleaning efficiency reached 96.8% with the optimized composition. Production trials confirmed that the cooling water pipe had no deformation or melt-through, and the average shear strength of the metallurgical bonding zone was 318 MPa.
Third, I developed an alcohol-based anti-sand coating using white corundum as the aggregate and an optimized combination of fluxes, suspending agents, and binders. The optimal flux ratio of potassium feldspar to iron oxide was 2:1, which gave excellent thermal shock resistance. Production validation demonstrated that the coating allowed more than 95% of the surface to be cleaned without excessive effort, greatly reducing sand foundry defects on the cast steel cooling wall.
Overall, my research provided effective solutions to the two key problems that previously limited the production and application of cast steel cooling walls. I believe that these techniques can be directly implemented in foundries to improve product quality, reduce manufacturing cost, and promote the wider use of cast steel cooling walls in modern blast furnaces.