In our production facility, we frequently encounter complex castings such as impellers and rotating discs for research and export orders. In 1995, we received a critical order for a brake rotor casting with an external diameter of 285 mm, a height of 122 mm, and 24 hollow blades each with an 8 mm cavity. The casting weighed 13.5 kg and was specified in alloy cast iron with the chemical composition shown in Table 1. The requirements demanded high dimensional accuracy, excellent surface finish, and absolute absence of shrinkage cavities, shrinkage porosity, or any other sand casting defect. The tensile strength had to exceed 276 MPa. Several foundries had previously attempted production but achieved a yield below 50%, making it impossible to fulfill the export contract. After we took over this task, we continuously improved the process and finally adopted a special casting technique, which raised the product yield to 98%–100% and ensured timely delivery.
| C | Si | Mn | Ni | Cr | Mo | Fe |
|---|---|---|---|---|---|---|
| ≥3.40 | 1.0–1.7 | 0.6–0.9 | 0.9–1.2 | 0.25–0.35 | 0.25–0.35 | Balance |
Casting Process Analysis and Scheme Determination
Molding Method
Given the high demands on dimensional accuracy, surface quality, and internal soundness, conventional green sand molding could not guarantee freedom from sand casting defect. We therefore adopted a combined mold technique using self-hardening resin sand together with spaced chills (sand-layered chills) to locally regulate cooling rates.
Process Improvements During Trials
In the initial trials we used a top-gating system with a shrink-bob riser. However, because the cavity shape was extremely complex, the metal flow became unstable during filling, leading to gas holes and cold shots (iron beads) in the upper disc region of the blades. After switching to a bottom-pour gating system, these defects were eliminated, but shrinkage porosity appeared at the location marked as B in the original sketch (we describe it as the area near the ingate). The cause was local overheating of the metal around the ingate due to prolonged high temperature. To address this, we placed spaced chills at that location to enhance cooling and equalize the temperature distribution. Unfortunately, the sand casting defect then shifted to other unchilled areas. Furthermore, a new shrinkage defect appeared at the upper part (position C) because the riser above did not provide adequate feeding to the hot spot. Enlarging the riser or adding a feed pad would have increased the machining allowance and production cost. Instead, we placed a ring-shaped spaced chill on the lower disc to increase the cooling capacity of the lower mold, and added local chills at the upper shrinkage area to eliminate the local hot spot. Figure 1 (inserted below) shows the final production process arrangement. Through these modifications, the product yield dramatically increased from 30% in the early stage to 98%, and in the first half of 1997 we produced nearly 200 pieces with 100% acceptance rate.
Chill and Gating/Riser System
The chills were made of cast iron, with holes (diameter 12–18 mm) either cast or drilled to allow the self-hardening sand to lock the chill in place and to enhance gas evacuation. A 3–3.5 mm layer of sand was retained between the chill and the casting, and the chills were positioned using a template during core and mold making.
The gating system was calculated using the well-known formula (often attributed to A.Z. or the “Atsu” equation) to determine the minimum choke area. For a choke cross-section area of the ingate, we obtained:
$$ F_{\text{inner}} = 287 \ \text{mm}^2 $$
The actual ingate area used in the die was:
$$ F_{\text{inner, actual}} = 19 \times 2 \times 8 = 304 \ \text{mm}^2 $$
The ratio among the sprue, runner, and ingate was maintained as:
$$ \frac{F_{\text{sprue}}}{F_{\text{runner}}}{F_{\text{ingate}}} = 2.3 : 2 : 1 $$
The structure of the gating system is represented schematically in the earlier design (similar to the configuration in the original text).
For the riser, we used a diameter 1.4 times the diameter of the hot spot circle (the modulus ratio). The riser size is given by:
$$ D_{\text{riser}} = 1.4 \times d_{\text{hot spot}} $$
where \(d_{\text{hot spot}}\) was determined from the casting cross-section.
Alloy Melting
The alloy was melted in a medium-frequency induction furnace following standard alloy cast iron procedures. We performed triple inoculation using Si-RE-Al (silicon, rare earth, aluminum) in the ladle. The melt was checked with a test sample before pouring. The tapping temperature was 1500–1580 °C, and the pouring temperature was 1380–1400 °C.
Key Process Characteristics
If conventional green sand molding had been used, the machining allowance would have been 4–8 mm, and the casting weight would have reached 24 kg. By adopting the self-hardening resin sand method with segmented molds and assembly, the dimensional accuracy improved significantly, allowing a machining allowance of only 2–2.5 mm, which reduced the casting weight to 15 kg. This resulted in substantial savings in material and machining costs, especially under mass production conditions.
The use of spaced chills allowed local adjustment of the cooling capacity of the mold, promoting progressive solidification (directional solidification) toward the riser without increasing the casting weight. This effectively eliminated shrinkage cavities and shrinkage porosity, which are common sand casting defect types in such complex geometries.
Practical Production Results
The process proved highly convenient for operators, as each step was visually manageable and adjustable regardless of batch size. The labor intensity was low, and the operators favored the method. Under mass production, both the cost and the cycle time were reduced while the quality improved. By strictly following the process regulations, we achieved a yield of 100% in many batches. Table 2 summarizes the improvement in yield over the course of development.
| Stage | Yield (%) | Notes |
|---|---|---|
| Initial top-gating | ~30 | Gas holes and cold shots |
| Bottom-pour without chills | ~50 | Shrinkage near ingate |
| With spaced chills (first version) | ~75 | Shrinkage shifted |
| Final optimized chills + riser | 98–100 | No sand casting defect |
Application of Similar Principles to Other Castings
We also applied comparable special techniques to other complex castings, such as a gearbox housing (pedestal) for a rolling mill. That casting had dimensions 2100×1950×2050 mm, weighed 18,500 kg (rough casting), with a pouring weight of 28,000 kg, material ZG35. The critical challenges included controlling solidification at heavy sections (maximum hot spot diameter 450 mm) and avoiding sand casting defect such as shrinkage, cracking, and misruns. We used a bottom-pour gating system with multiple ingates (three levels) to distribute the metal evenly, combined with a combination of internal and external chills. The gating ratios were:
$$ F_{\text{pouring cup}} : \Sigma F_{\text{sprue}} : \Sigma F_{\text{runner}} : \Sigma F_{\text{ingate}} = 1 : 2 : 2.25 : 2 $$
The pouring time was calculated as:
$$ t = \frac{Q}{n \cdot q} = \frac{18500}{1 \times 150} = 123.33 \ \text{s} $$
where \(Q\) is the pouring weight (kg), \(n\) is the number of ladles, and \(q\) is the pouring rate per ladle (kg/s). The rise velocity of the molten steel in the mold cavity was:
$$ V = \frac{H}{t} = \frac{2050}{123.33} = 16.62 \ \text{mm/s} $$
This velocity was adequate to avoid cold shuts and other sand casting defect.
The risers were open-top cylindrical risers, each with a volume calculated using the modulus method. For a hot spot diameter of 450 mm, we used a riser of 500 mm diameter and 1500 mm height. Internal chills were made from 20 mm diameter round steel bars welded into a grid, totaling 350 kg (5% of the steel weight in the hot zone). External chills were placed at all T-junctions and cross sections to reduce hot spots and prevent cracks.
To prevent distortion of the U-shaped opening, we added four tie bars (two with dimensions 150×250 mm and two with 30×120 mm). Contraction ribs were added at the junction of the 50 mm and 350 mm walls to prevent hot tearing. The ribs had a thickness of 13 mm and a spacing of 20 mm.
The long hole (diameter 120 mm, length 2100 mm) was cast using a seamless steel tube (wall thickness 15 mm) filled with molding sand. Steel bars were welded to the tube surface to prevent melting and distortion, and the assembly was pickled before being embedded in the mold.
Summary of Advantages
The special casting technology described here—combining self-hardening resin sand, spaced chills, carefully designed gating and risering, and appropriate use of internal/external chills—effectively eliminated common sand casting defect such as gas porosity, cold shots, shrinkage cavities, and hot tears. Table 3 summarizes the key parameters used in the impeller disc casting.
| Parameter | Value |
|---|---|
| Mold material | Self-hardening resin sand + spaced chills |
| Gating system type | Bottom-pour with multi-gates |
| Ingate cross-section (total) | 304 mm² |
| Riser diameter factor | 1.4 × hot spot diameter |
| Chill material | Cast iron with holes for sand locking |
| Sand layer between chill and casting | 3–3.5 mm |
| Pouring temperature | 1380–1400 °C |
| Final yield | 98–100% |
We believe that for optimal production of impeller and rotating disc castings, this process offers visual control of every step, easy adjustment of the mold structure, and reliable quality control. By adhering strictly to the process, a sand casting defect rate of zero is achievable.