In 2005, our factory was tasked with producing a high-speed elevator thin-motor rotor seat made of ductile iron (grade QT450-10) for a Japanese client. The casting had critical dimensional specifications: a contour size of φ560×190 mm, a main wall thickness of 45 mm, and a net weight of 97 kg. As a high-speed rotating component, the casting required full machining followed by dynamic and static balancing tests. Any shrinkage cavities, porosity, or cracks were strictly prohibited. Critical surfaces tolerated zero gas porosity or sand inclusion defects, while other surfaces allowed only two defects measuring no more than 1.5×1×1 mm. Sand casting defects of any kind, especially shrinkage, were unacceptable. This article details our journey to overcome persistent shrinkage defects through systematic process modifications, emphasizing the critical role of feeder design and green sand mold behavior in preventing sand casting defects.
Production Background
Our gray iron foundry primarily produced engine cylinder blocks for various models. The molding line employed an air-impact molding process with green sand, using flask dimensions of 1200×800×350 mm. Melting was performed in a 6-ton medium-frequency induction furnace. The challenge was unprecedented because we had never produced ductile iron castings on this high-pressure green sand line. The primary concern was preventing shrinkage porosity and cavities, as ductile iron exhibits significant graphitization expansion during solidification. Proper mold rigidity is essential to harness this “self-feeding” effect. However, green sand molds can deform under metallostatic pressure and expansion forces, exacerbating sand casting defects like shrinkage. Additionally, high moisture content in green sand can cause subsurface pinholes and slag inclusions. Many domestic foundries used no-bake sand systems with riserless casting for medium-section ductile iron, but we had to adapt to our existing green sand line, making sand casting defects a major risk.
Casting Analysis and Initial Approach
The casting geometry presents a stepped section with a heavy central hub (φ290) and a thin outer rim (φ480). The critical shrinkage-prone areas were identified at the root fillets of φ480 and φ290, as indicated by machining experience. The initial design followed the “balanced solidification” theory, aiming for simultaneous solidification through multiple ingates and riserless casting. Given the high surface hardness of air-impact molded green sand (measured 75–95 on a wet hardness tester) and its chilling effect, we believed that with proper gating and chills, we could avoid external feeders. The first attempt used a bottom-gating system with six ingates arranged around the casting, plus four overflow risers on top, as shown conceptually in Figure 1 (the image link below corresponds to typical sand casting defects encountered).
This image illustrates typical sand casting defects that we strove to avoid, particularly the shrinkage cavities visible on machined surfaces.
Step 1: Riserless Bottom-Gating with Chills
The first production trial (Plan 1) employed a bottom-gating system with six uniform ingates. No risers were used, based on the “self-feeding” concept. The casting surface appeared sound, but after machining, visible shrinkage porosity appeared along approximately one-third of the arc at the root of φ480 and φ290. To counter this, we added two rows of external chills (ring-shaped) around these radii. Subsequent trials eliminated the shrinkage, but the use of chills introduced severe production issues:
- The high-speed automated molding line could not pause for chill placement without reducing cycle time. Each chill required separate casting, cleaning, grinding, rust removal, welding, coating, and drying—a logistical nightmare for mass production.
- Chills increased the rejection rate due to misplacement or gaps between chills, causing localized shrinkage anyway.
- Chills also promoted subsurface gas porosity on the top surface, increasing scrap.
Because of these drawbacks, we abandoned the chill approach and sought a more robust solution to sand casting defects without external cooling aids.
Step 2: Top-Gating with Hot and Cold Riser Combinations
We redesigned the process (Plan 2) to top-gating by inverting the cope and drag flasks. This allowed larger riser volumes at the top, enhancing feeding pressure. A ceramic foam filter was placed below the sprue to reduce slag and sand inclusions. The gating system followed the “large orifice flow” and “balanced solidification” principles, using thin and wide ingates. Four “pressure-edge” risers (also called kissing risers) were arranged: two of them served as both ingate and riser (hot risers, through which iron entered the cavity), while the other two acted as overflow risers (cold risers, receiving only cold iron after the cavity filled). The kiss width was 6 mm. Table 1 summarizes the key parameters of Plan 2.
| Parameter | Value |
|---|---|
| Pouring method | Top-gating via two hot risers |
| Hot riser dimensions (mm) | φ50×60 (height) |
| Cold riser dimensions (mm) | φ50×60 (height) |
| Kiss width (mm) | 6 |
| Ingate thickness (mm) | 4 (wide and thin) |
| Number of overflow risers | 2 (cold) |
| Number of hot risers | 2 |
| Filter location | Below sprue, 10 ppi foam |
| Sand mold hardness | 85–95 (wet hardness) |
Castings from Plan 2 showed no surface defects like gas holes or slag inclusions. However, machining revealed concentrated shrinkage cavities (2–3 spots) and severe porosity at the φ480 bottom fillet, precisely opposite the hot riser and ingate locations. We analyzed the root cause:
- Despite high-pressure molding, green sand molds still deform under thermal expansion, causing cavity enlargement during solidification. The graphitization expansion in ductile iron is beneficial only if the mold cavity does not grow; otherwise, it creates a feeding demand that must be satisfied.
- The hot riser’s kiss width (6 mm) was too small. It solidified prematurely, blocking the liquid feeding channel before the casting had finished solidifying. Meanwhile, the cold riser, even though thicker and positioned higher, contained cooler iron that froze even earlier than the casting. Instead of feeding, it actually extracted liquid from the casting, aggravating sand casting defects.
- The ingate thickness was too thin, freezing quickly and isolating the riser from the casting, preventing any effective feeding during the later stages of solidification.
These observations invalidated the assumption that “cold risers” could provide any feeding benefit. They only acted as vents or slag traps, but their early solidification made them detrimental.
Step 3: Optimized Hot Riser with Enlarged Kiss and Thin Ingates
Based on the failure analysis, we implemented a third design (Plan 3) with the following modifications:
- Eliminated both cold overflow risers entirely. Only six small vent holes (φ8 mm) were left for gas escape.
- Increased the kiss width of the two hot risers from 6 mm to 20 mm. The riser diameter was also enlarged to φ60×70 mm height. The enlarged kiss ensured that the liquid feeding channel remained open long enough to compensate for the casting’s contraction during the entire solidification period.
- Changed the ingate design: instead of a thin gate, we used a thin and wide gate (10 mm thick × 40 mm wide) that would solidify quickly after pouring, effectively sealing the riser-casting junction at the right moment. This forced the riser to become the last solidifying part of the casting-riser system, shifting any final shrinkage into the riser.
- Maintained top-gating and ceramic foam filtration.
Table 2 compares the critical parameters across the three plans.
| Parameter | Plan 1 (Riserless) | Plan 2 (Hot+Cold) | Plan 3 (Optimized Hot) |
|---|---|---|---|
| Gating type | Bottom | Top via hot risers | Top via hot risers |
| Number of risers | 0 (overflow only) | 2 hot + 2 cold | 2 hot |
| Riser diameter (mm) | N/A | 50 | 60 |
| Riser height (mm) | N/A | 60 | 70 |
| Kiss width (mm) | N/A | 6 | 20 |
| Ingate thickness (mm) | 6 (4 ingates) | 4 | 10 |
| Ingate width (mm) | 40 | 50 | 40 |
| Chills used | Yes (later removed) | No | No |
| Overflow risers | 4 (top) | 2 cold | 0 |
| Filter | None | Ceramic foam | Ceramic foam |
| Result (shrinkage) | Visible at fillets | Concentrated at φ480 | None detected |
The theoretical basis for the enlarged kiss can be expressed by the feeding distance and riser modulus. For ductile iron, the modulus of the riser \( M_R \) must exceed the modulus of the casting section \( M_C \) by a safety factor, often 1.1 to 1.2. The modulus is defined as the volume-to-cooling surface area ratio:
$$ M = \frac{V}{A} $$
For the critical section at the rim (φ480 fillet), the local modulus was estimated as 1.8 cm. The initial hot riser (φ50×60) had a modulus of approx. 1.6 cm, insufficient. The enlarged riser (φ60×70) gave a modulus of 2.0 cm, providing a 1.11 ratio. Additionally, the kiss width governs the heat flow from riser to casting. The feeding channel must remain liquid until the casting has completed its solidification shrinkage. The time for the kiss to solidify, \( t_k \), can be approximated by Chvorinov’s rule:
$$ t_k = C_m \left( \frac{M_k}{C} \right)^2 $$
where \( M_k \) is the modulus of the kiss, \( C_m \) is the mold constant, and \( C \) is a constant for the mold material. A kiss width of 20 mm gives a modulus of 1.0 cm (considering the kiss as a short cylinder), compared to 0.3 cm for the 6 mm kiss. This increased \( t_k \) significantly, allowing the riser to feed longer.
We conducted three production runs with Plan 3. Four castings were randomly selected for destructive sectioning. No shrinkage cavities or porosity were found. Subsequently, 18 castings were machined and passed the client’s stringent inspection, including dynamic balancing. The sand casting defects were completely eliminated.
Discussion: Key Lessons for Preventing Sand Casting Defects in Ductile Iron via Green Sand Molding
Our experience highlights several fundamental principles for avoiding sand casting defects, especially shrinkage, in ductile iron produced on high-pressure green sand lines:
- Sequence solidification is essential. Ductile iron exhibits a relatively narrow freezing range and a large graphitization expansion. However, green sand molds, even when highly compacted, cannot completely resist the expansion. The mold cavity expands during solidification, reducing the available expansion pressure. Therefore, relying solely on “self-feeding” is risky for medium-section castings. External feeding with properly sized hot risers is more reliable.
- Hot risers must be large enough and have a sufficiently wide connection to the casting. The kiss width should be at least one-third of the adjacent casting wall thickness, and the riser modulus should exceed that of the casting section by ≥10%. A narrow kiss solidifies prematurely, creating a “choke” that prevents feeding and actually makes sand casting defects worse.
- Cold risers (those not receiving hot metal) are completely ineffective for feeding and can be harmful. Their thermal lag means they solidify before the casting, so they cannot provide any liquid metal. At best, they act as vents; at worst, they can promote shrinkage by acting as heat sinks that accelerate local solidification. We recommend eliminating them entirely and using only small vents for gas escape.
- Ingate design should balance fast sealing with adequate feeding. A thin, wide ingate solidifies quickly after pouring, isolating the riser and forcing the casting to use its own expansion. However, the ingate must remain open long enough to allow the riser to feed the initial contraction. In our optimized design, the ingate thickness (10 mm) was still significantly less than the kiss width (20 mm), ensuring that the ingate sealed before the kiss, allowing the riser to feed through the kiss until the very end.
- Top-gating via the riser is beneficial. It establishes a thermal gradient that favors the riser as the hottest part, maintaining a positive feeding path. It also provides higher metallostatic pressure at the riser base.
- Avoid chills in automated green sand lines. Chills are labor-intensive and can cause other sand casting defects like gas porosity. With proper riser design, chills are unnecessary for this casting geometry.
Conclusions
Through a systematic trial-and-error approach, we successfully eliminated shrinkage defects in a medium-section ductile iron rotor seat casting produced on a high-pressure green sand molding line. The critical changes were:
- Replacing bottom-gating with top-gating through two hot risers.
- Enlarging the hot riser kiss width from 6 mm to 20 mm, and increasing the riser size to achieve a modulus ratio above 1.1.
- Designing thin, wide ingates that solidify promptly after filling, sealing the riser-casting junction while the riser remains liquid via the wide kiss.
- Removing all cold overflow risers, which were counterproductive.
These modifications ensured that the casting received adequate liquid metal throughout its solidification, preventing the formation of sand casting defects such as shrinkage cavities and porosity. The process has been proven in mass production with zero internal defects over multiple batches. The key to preventing sand casting defects in ductile iron green sand casting lies in deliberate, directional solidification with robust hot risers, rather than relying on mold rigidity or chills alone.
Summary of Parameters and Results
| Parameter | Value | Remark |
|---|---|---|
| Casting weight (kg) | 97 | |
| Main wall thickness (mm) | 45 | |
| Riser modulus (cm) | 2.0 | Sufficient > 1.1 × 1.8 |
| Kiss modulus (cm) | 1.0 | Provided feeding time ~ 3× longer than Plan 2 |
| Pouring temperature (°C) | 1,380–1,420 | |
| Mold hardness (wet) | 85–95 | |
| Scrap rate from sand casting defects | 0% | After optimization |
| Number of castings inspected | 22 (4 sectioned + 18 machined) | All passed |
In summary, our work demonstrates that even with the inherent limitations of green sand molds, sand casting defects can be fully controlled by applying sound risering principles tailored to ductile iron’s solidification characteristics. The final process is now considered standard for this family of rotor seat castings.