In our foundry, two new impact molding lines were put into production for manufacturing track shoe castings. Soon after commissioning, we encountered a severe sand casting defects problem: shrinkage cracks appearing in a high percentage of the castings. The defect rate was alarmingly high, as shown in the initial statistics. Over a production period, the total inspected castings were 78,148 pieces, with 18,414 pieces being defective, yielding a scrap rate of 23.57%. Among these, shrinkage cracks accounted for 14.30%, while other defects like underfill and composition issues contributed 1.40% and 6.00% respectively. This situation necessitated a thorough investigation into the root causes of these sand casting defects and the development of effective countermeasures.
| Inspected pieces | Total defective | Scrap rate (%) | Shrinkage crack rate (%) | Underfill rate (%) | Composition scrap rate (%) |
|---|---|---|---|---|---|
| 78,148 | 18,414 | 23.57 | 14.30 | 1.40 | 6.00 |
We observed that the shrinkage crack defects exhibited several distinct characteristics. First, the quantity was substantial. Second, the defect rate was significantly higher than that on the older Z148B molding line. Third, the cracks were consistently located at the junction between the guide rib and the rolling surface, specifically at the center of the corner area. Fourth, the crack morphology was consistent: width less than 3 mm and length within 30 mm. Fifth, the crack occurrence rate fluctuated widely, ranging from 1% to 50% across different batches. Lastly, when we sectioned the defective castings perpendicular to the crack, we found evidence of shrinkage porosity beneath the surface, indicating that the cracks were indeed initiated by shrinkage.
To understand the formation mechanism of these sand casting defects, we considered two major aspects. First, the casting design itself might contain oversized hot spots or unfavorable geometries for heat dissipation. Second, the impact molding process differed from the Z148B line in several key factors, including melting practice, pouring temperature, inclusion content, slag basicity, and most importantly, the molding method which resulted in significantly different mold hardness distributions and thus different restraint forces on the solidifying casting.
2. Influence of Casting Geometry
We first examined the pattern plates for the impact molding line. It was found that the hot spots on the patterns were generally oversized due to pattern manufacturing issues. We modified three critical areas: a prominent hot spot at position 1, the area corresponding to the shrinkage crack location at position 2, and the removal of a feeder channel at position 3 (where the ingate used to be). After these modifications, the crack length decreased from over 30 mm to less than 20 mm, but the overall scrap rate did not drop significantly.
Later, in 1995, the casting design for track shoe part number 75.34.101B was revised to solve a tractor derailment problem. The new design, 75.34.101C, featured a higher guide rib (increased by 11 mm) and some structural changes. The most impactful change was the geometry at the corner where the guide rib meets the rolling surface. The old design (75.34.101B) had a shallow and flat depression underneath, which promoted heat dissipation. The new design (75.34.101C) had a deep and steep depression, which hindered heat flow. Consequently, the effective hot spot size changed: for the old design it was approximately 13.6 mm in diameter, while for the new design it was 17.3 mm. This increase in hot spot size made the new design more prone to shrinkage cracking.
We performed a comparative experiment by mounting both designs on the same pattern plate: one piece of the old design and three pieces of the new design. Two casting trials were conducted: top-pouring and bottom-pouring. The results for mixed top-pouring are summarized in the following table.
| Part number | Total pieces | Shrinkage cracked pieces | Crack length (mm) | Crack rate (%) |
|---|---|---|---|---|
| 75.34.101B (old) | 42 | 16 | ≤5 | 38.1 |
| 75.34.101C (new) | 126 | 54 | 10–15 | 42.9 |
The old design still showed a high crack rate of 38.1%, although the cracks were much smaller. The new design had a slightly higher crack rate (42.9%) and longer cracks. We then performed mixed bottom-pouring tests and compared with simultaneous top-pouring results for the new design.
| Part number & pouring method | Total pieces | Shrinkage cracked pieces | Shrinkage crack rate (%) | Underfill pieces | Underfill rate (%) |
|---|---|---|---|---|---|
| 75.34.101B bottom-pouring | 105 | 5 | 4.7 | 1 | 0.98 |
| 75.34.101C bottom-pouring | 383 | 40 | 10.4 | 8 | 2.1 |
| 75.34.101C top-pouring | 667 | 88 | 13.2 | 78 | 11.7 |
Bottom-pouring significantly reduced both shrinkage cracks and underfill defects for the new design. The old design with bottom-pouring achieved a crack rate of only 4.7%, which was promising but still not entirely satisfactory. The results confirmed that reducing hot spot size and improving heat dissipation conditions are crucial for mitigating sand casting defects related to shrinkage.
3. Differences Between Impact Molding and Z148B Lines
We investigated the effect of pouring temperature on shrinkage crack formation. During a typical heat, pouring temperature gradually decreased over 30–40 minutes. We divided the pouring sequence into three segments: initial, middle, and final. The scrap rate for each segment was recorded using the unmodified 75.34.101C top-pouring pattern. The results are shown below.
| Segment | Total pieces | Shrinkage cracked pieces | Shrinkage crack rate (%) |
|---|---|---|---|
| Initial | 150 | 23 | 15.3 |
| Middle | 150 | 24 | 16.0 |
| Final | 148 | 21 | 14.2 |
The variation was insignificant, indicating that pouring temperature within the normal range did not play a dominant role in causing these sand casting defects. We also tried adding scrap steel to the ladle to control temperature, but no improvement was observed.
Next, we tested the influence of steel cleanliness and slag basicity by using molten steel from the electric furnace that was normally used for the Z148B line and pouring it into the impact molding line. The results were striking.
| Total pieces | Shrinkage cracked pieces | Shrinkage crack rate (%) | Underfill pieces | Underfill rate (%) |
|---|---|---|---|---|
| 195 | 64 | 32.8 | 11 | 5.6 |
Both shrinkage and underfill rates were very high, similar to the levels observed with the impact line’s own steel. This suggested that the root cause was not metallurgical quality but rather the molding process itself.
We then measured the mold hardness on both lines. The average values are given in the following table.
| Molding line | Top mold hardness | Bottom mold hardness |
|---|---|---|
| Z148B | 78 | 80 |
| Impact molding | 91 | 90 |
The impact molding line produced molds with significantly higher hardness (91/90) compared to the Z148B line (78/80). High mold hardness means low collapsibility and high resistance to casting contraction during solidification. This mechanical restraint is a major contributor to the formation of shrinkage cracks, which are a common type of sand casting defects.
Analyzing the casting geometry, the guide rib is the main source of contraction resistance. The pattern layout shows that one side of the track shoe is connected by the center pin ear, and the other side is connected by the runner bar. Intuitively, the guide rib should experience a contraction force in the B direction (compressive), which would help prevent cracks. However, after measuring the castings, we found that the guide rib actually tilted inward by about 1°, indicating that the contraction resistance acted in the A direction (tensile). By examining the temperature field after pouring and the solidification sequence, we identified the source of this resistance. The runner bar, being the passage for molten steel, remained at a much higher temperature than the casting and had a larger cross-section. The solidification order was: guide rib → other parts of the casting → center pin ear → runner bar. When the guide rib and other parts contracted, the runner bar and center pin ear were still hot and weak. Because the impact-molded sand had very high hardness, it strongly resisted the contraction of the guide rib, creating tensile stresses that caused the guide rib to tilt inward and eventually crack. Later, when the runner bar finally contracted, the casting was already rigid, and the internal stresses sometimes caused breakage at the ingate.
4. Countermeasures to Reduce Shrinkage Cracks
To mitigate these sand casting defects, we adopted two main strategies: improving sand collapsibility and optimizing casting design. Reducing mold hardness directly on the impact molding machine is difficult due to equipment limitations. Therefore, we introduced a sand additive (DS-type additive) that enhances mold collapsibility. Initially, we added 1% DS additive to the new sand replenishment. After 4–5 days when the system became uniform, we tracked the casting quality.
| Total pieces inspected | Shrinkage cracked pieces | Shrinkage crack rate (%) |
|---|---|---|
| 583 | 20 | 3.43 |
The crack rate dropped dramatically to 3.43%. However, after one week without adding DS additive, the rate rebounded.
| Total pieces inspected | Shrinkage cracked pieces | Shrinkage crack rate (%) |
|---|---|---|
| 415 | 26 | 6.3 |
This confirmed that sand collapsibility was a key factor. We then optimized the process by adding 0.5%–1.0% DS additive to the new sand and also adding 5–10 kg of DS additive per batch of bentonite slurry. This brought the shrinkage crack rate down to about 1%.
Furthermore, we redesigned the casting structure by changing the orientation of the reinforcement ribs on the back side from transverse to longitudinal, which reduced the hot spot size. We modified two out of four patterns on a plate, leaving two unchanged for comparison. With the DS additive continuously used, the longitudinal-rib castings showed zero shrinkage cracks over six months of production, while the unchanged transverse-rib castings still showed about 1% scrap.
5. Theoretical Analysis of Shrinkage Crack Formation
To quantify the influence of geometry and sand restraint, we employed solidification modulus and stress calculations. The shrinkage crack formation involves a competition between the tensile stress generated by contraction and the high-temperature strength of the casting. The tensile stress can be approximated as:
$$ \sigma = E \cdot \alpha \cdot \Delta T – \sigma_{yield} $$
where \(E\) is the elastic modulus, \(\alpha\) is the thermal expansion coefficient, \(\Delta T\) is the temperature range over which contraction occurs, and \(\sigma_{yield}\) is the yield strength at high temperature. When the sand mold offers high resistance, the effective contraction is constrained, leading to higher stress. The resistance force from the sand can be expressed as:
$$ F_{res} = k \cdot A \cdot H $$
where \(k\) is a constant related to sand stiffness, \(A\) is the contact area, and \(H\) is the mold hardness. Since the impact molding line had \(H \approx 91\) versus Z148B’s \(H \approx 79\), the resistance was approximately \((91/79)^2 \approx 1.33\) times higher (assuming linear relationship with hardness squared). This increase in resistance directly raises the likelihood of sand casting defects like shrinkage cracks.
The hot spot size can be characterized by the local modulus \(M = V/A\), where \(V\) is the volume and \(A\) is the cooling surface area. For the old design (75.34.101B), the modulus at the critical corner was smaller because the depression was shallow. Let \(M_{old} \approx 13.6/4 = 3.4\) mm (assuming cylindrical geometry), while for the new design \(M_{new} \approx 17.3/4 = 4.3\) mm. The solidification time is proportional to the square of the modulus:
$$ t_{solid} \propto M^2 $$
Thus the new design’s solidification time was about \((4.3/3.4)^2 \approx 1.6\) times longer. This longer time allows more contraction to occur before the surrounding structure gains strength, exacerbating the crack tendency.
We also developed an empirical relationship between scrap rate \(S\) (in percent) and the ratio of resistance to casting strength:
$$ S = a \cdot \left( \frac{F_{res}}{F_{strength}} \right)^b $$
where \(a\) and \(b\) are constants. From our data, using the impact line before additive, the resistance was high and scrap rate was ~14%. After additive, resistance dropped and scrap rate fell to ~1%. This relationship, though simplified, helps guide future optimization.
6. Long-Term Results and Discussion
After implementing the combination of longitudinal ribs and DS additive, we monitored the production for over six months. The shrinkage crack rate stabilized at below 1% for the modified design, while the unmodified design (with rib orientation unchanged but still using DS additive) remained around 1% scrap. The results are summarized in the following table.
| Design | Sand additive | Total pieces produced | Shrinkage cracked pieces | Shrinkage crack rate (%) | Other defects rate (%) |
|---|---|---|---|---|---|
| Longitudinal ribs | 0.5%–1.0% DS | ~12,000 | 0 | 0 | ~2.0 |
| Transverse ribs (original) | 0.5%–1.0% DS | ~12,000 | ~120 | ~1.0 | ~2.5 |
These results confirm that the combination of improved sand collapsibility and optimized casting geometry is highly effective in eliminating sand casting defects due to shrinkage. The cost of adding DS additive is modest, and the redesign of ribs did not compromise the functional requirements of the track shoe. In fact, the longitudinal ribs provided even better strength distribution.
7. Conclusions
Through systematic analysis and experimental verification, we reached the following conclusions regarding the shrinkage crack sand casting defects in impact-molded track shoe castings:
1. Pouring temperature within the normal operational range did not significantly affect the crack rate. The primary causes were high mold hardness (low collapsibility) and oversized local hot spots in the casting design.
2. The introduction of 0.5%–1.0% DS-type sand additive to the molding sand effectively improved mold collapsibility, reducing the mechanical restraint during solidification. This alone lowered the crack rate from over 10% to about 1%–3%.
3. Changing the reinforcement ribs from transverse to longitudinal orientation reduced the hot spot modulus, further decreasing the shrinkage crack tendency. The combination of longitudinal ribs and DS additive eliminated the defect entirely for the modified castings.
4. The theoretical framework using solidification modulus and stress resistance models helped explain the observed behavior and can be used for predictive optimization of other casting designs prone to similar sand casting defects.
These findings have been successfully applied to our production, significantly improving yield and reducing costs. The lessons learned are applicable to other ferrous castings produced in high-pressure molding lines where sand collapsibility is limited.