We have encountered a severe sand casting defect in the form of shrinkage cracks in crawler belt castings produced on our newly installed air-impact molding lines. This sand casting defect, with a defect rate exceeding 23%, has significantly impacted production efficiency and investment benefits. Through systematic investigation, we identified the root causes and implemented effective countermeasures to mitigate this sand casting defect. In this paper, we present our findings, experimental data, and solutions in a comprehensive manner, emphasizing the role of casting structure, mold rigidity, and sand collapsibility in controlling sand casting defect occurrence.
The analysis focused on several factors potentially leading to the sand casting defect: pouring temperature, initial temperature field of the casting, resistance during solidification shrinkage, sand mold collapsibility, and casting geometry. Based on these factors, we proposed structural modifications (changing the rib orientation) and the addition of a special sand additive (DS additive) to improve mold collapsibility. The results demonstrated a drastic reduction in the sand casting defect rate from over 14% to below 1%.
To provide a clear context, the following image illustrates typical sand casting defects including shrinkage cracks:
1. Introduction
Our foundry operates two air-impact molding lines for producing crawler belt castings (material: low-alloy steel). Shortly after commissioning, we observed a high proportion of scrap due to a specific sand casting defect—shrinkage cracks located at the intersection of the guide rib and the running surface. Table 1 summarizes the scrap statistics over a production period.
| Inspected (pcs) | Scrap (pcs) | Scrap rate (%) | Shrinkage crack rate (%) | Misrun rate (%) | Composition reject rate (%) |
|---|---|---|---|---|---|
| 78,148 | 18,414 | 23.57 | 14.30 | 1.40 | 6.00 |
The shrinkage crack defect dominated the scrap, and its occurrence was highly inconsistent, fluctuating from 1% to 50% between heats. This sand casting defect displayed consistent features: narrow cracks (width < 3 mm, length ≤ 30 mm) at a fixed location, and cross-section inspection revealed internal shrinkage porosity beneath the crack. These characteristics suggested a combination of thermal and mechanical factors during solidification.
2. Defect Characteristics
We systematically recorded the morphology and location of this sand casting defect. The crack always appeared at the interior angle between the guide rib and the running surface. In contrast, the same design produced on our older Z148B molding lines rarely exhibited such a defect. This disparity directed our attention to the differences between the two molding processes.
3. Root Cause Analysis
3.1 Casting Structure
We first examined the pattern for the air-impact line and found that the local hot spots were larger than those on the Z148B pattern due to pattern manufacturing inaccuracies. Figure 2 (not shown here) indicates three areas of excessive metal thickness. We reduced these sections: area 1 was a distinct hot spot; area 2 was located directly above the crack site; area 3 removed a runner that acted as a feeder. After modification, the crack length reduced from >30 mm to <20 mm, but the overall defect rate did not improve significantly.
In 1995, the casting design changed from part number 75.34.101B to 75.34.101C to prevent track derailment. The new design increased the guide rib height by 11 mm and altered the geometry at the rib–surface junction. This change enlarged the local thermal modulus. For 75.34.101B, the equivalent hot-spot diameter was approximately 13.6 mm, while for 75.34.101C it increased to 17.3 mm. A larger hot spot promotes a sand casting defect because it delays solidification and creates a vulnerable liquid film under tensile stress. We conducted mixed-pattern trials (one cavity of 75.34.101B and three cavities of 75.34.101C on the same plate) using both top-pouring and bottom-pouring configurations.
| Part No. | Total (pcs) | Cracked (pcs) | Crack length (mm) | Crack rate (%) |
|---|---|---|---|---|
| 75.34.101B | 42 | 16 | ≤5 | 38.1 |
| 75.34.101C | 126 | 54 | 10–15 | 42.9 |
| Part No. & method | Total (pcs) | Shrinkage crack (pcs) | Crack rate (%) | Misrun (pcs) | Misrun rate (%) |
|---|---|---|---|---|---|
| Bottom-pour 75.34.101B | 105 | 5 | 4.7 | 1 | 0.98 |
| Bottom-pour 75.34.101C | 383 | 40 | 10.4 | 8 | 2.1 |
| Top-pour 75.34.101C | 667 | 88 | 13.2 | 78 | 11.7 |
Bottom-pouring (casting upside down) improved the thermal gradient and reduced misruns, but the sand casting defect rate for 75.34.101C remained at 10.4%—still unacceptable. The older design 75.34.101B with bottom-pouring achieved 4.7%, indicating that geometry plays a crucial role in this sand casting defect.
3.2 Molding Process Differences: Air-Impact vs. Z148B
The critical distinction between the two lines is mold hardness. Table 4 compares the average mold surface hardness (measured with a standard penetrometer).
| Molding method | Top mold (units) | Bottom mold (units) |
|---|---|---|
| Z148B | 78 | 80 |
| Air-impact | 91 | 90 |
The air-impact line produces molds with hardness values around 90, whereas the Z148B line produces softer molds (~78-80). Higher mold hardness increases the resistance to casting contraction, especially when the casting is restrained by sand cores and runners. To isolate the effect of molten steel quality, we transported steel melted for the Z148B line to the air-impact line and poured it into air-impact molds. The results (Table 5) showed similarly high defect rates, confirming that the sand casting defect is not caused by steel composition or superheat.
| Total (pcs) | Shrinkage crack (pcs) | Crack rate (%) | Misrun (pcs) | Misrun rate (%) |
|---|---|---|---|---|
| 195 | 64 | 32.8 | 11 | 5.6 |
We also investigated the influence of pouring temperature. During a single heat, the pouring temperature drops by about 30–40°C from start to end. We divided one heat into three segments and recorded the defect rate for each (Table 6). No significant correlation was found.
| Segment | Total (pcs) | Cracked (pcs) | Crack rate (%) |
|---|---|---|---|
| Start | 150 | 23 | 15.3 |
| Middle | 150 | 24 | 16.0 |
| End | 148 | 21 | 14.2 |
Thus, temperature is not a primary driver of this sand casting defect.
3.3 Shrinkage Resistance and Solidification Sequence
Analyzing the casting geometry and mold restraint, we identified the solidification sequence: the guide rib solidifies first, followed by the main body, then the middle pin ears, and finally the runner bar. The runner bar, being the hottest and largest section, remains liquid or mushy while the casting contracts. Because the air-impact mold is extremely hard, the contracting guide rib experiences a strong restraining force in the direction toward the runner. This force causes the rib to tilt inward (approximately 1° tilt measured on actual castings), creating tensile stresses at the fillet where the crack initiates. The stress distribution can be approximated by a simple thermal stress model.
Let the contraction strain be:
$$ \varepsilon = \alpha \cdot \Delta T $$
where α is the coefficient of thermal expansion (~12×10⁻⁶ /°C for steel) and ΔT is the temperature drop from solidus to room temperature (~1400°C). The elastic modulus E at high temperature is around 20 GPa. The theoretical thermal stress if fully restrained is:
$$ \sigma = E \cdot \varepsilon = 20 \times 10^9 \times 12 \times 10^{-6} \times 1400 = 336 \, \text{MPa} $$
This far exceeds the hot strength of the steel, causing hot tearing. In practice, some relaxation occurs, but the high mold hardness restricts free contraction, promoting this sand casting defect.
We also measured the local thermal modulus at the crack site. For the 75.34.101C design, the equivalent hot-spot diameter can be calculated from geometry:
$$ d_{eq} = \frac{4 \cdot (\text{section area})}{\text{perimeter}} $$
For the 75.34.101C fillet, deq ≈ 17.3 mm, while for 75.34.101B, deq ≈ 13.6 mm. The larger modulus increases solidification time and the risk of a sand casting defect.
4. Countermeasures
4.1 Reducing Mold Restraint: DS Sand Additive
To improve mold collapsibility without major equipment changes, we introduced a commercial sand additive (DS-type) that burns out during casting, creating voids and reducing the effective mold stiffness. Initially, we added 1% DS additive to the new sand make-up. After 4–5 days to stabilize the sand system, we tracked the defect rate (Table 7).
| Total inspected (pcs) | Cracked (pcs) | Crack rate (%) |
|---|---|---|
| 583 | 20 | 3.43 |
After stopping the additive for one week, the defect rate rebounded (Table 8), confirming the additive’s effectiveness.
| Total inspected (pcs) | Cracked (pcs) | Crack rate (%) |
|---|---|---|
| 415 | 26 | 6.3 |
We then optimized the addition method: 0.5%–1.0% in the make-up sand plus 5–10 kg per batch of bentonite slurry. This consistently reduced the sand casting defect to about 1%.
4.2 Structural Modification: Longitudinal Ribs
To further reduce the thermal modulus at the critical location, we changed the transverse reinforcement ribs on the back of the casting to longitudinal ribs (aligned with the casting’s long axis). This eliminated the concentrated hot spot at the rib–surface junction. We modified two cavities out of four on a pattern plate, leaving two cavities unchanged for comparison. After implementing the DS additive simultaneously, the longitudinal-rib cavities exhibited zero shrinkage cracks over six months of production, while the unchanged cavities still showed ~1% defect rate. This confirms that combining structural optimization with enhanced mold collapsibility is the most effective strategy against this sand casting defect.
5. Results and Discussion
Table 9 summarizes the final defect rates achieved with the combined countermeasure.
| Configuration | Total production (approx. pcs) | Shrinkage crack rate (%) | Misrun rate (%) |
|---|---|---|---|
| Longitudinal ribs (modified) | >10,000 | 0 | <0.5 |
| Transverse ribs (original) | >10,000 | ~1.0 | <0.5 |
The longitudinal rib design reduces the local thermal modulus, as quantified by:
$$ d_{eq,\text{long}} = \frac{4A_{\text{long}}}{P_{\text{long}}} \approx 11.2 \,\text{mm} $$
compared to 17.3 mm for the transverse rib. The smaller modulus shortens solidification time and minimizes the vulnerable period when hot tearing can occur. Additionally, the DS additive lowers the mold’s effective stiffness, reducing the restraint stress. The combined effect eliminates the sand casting defect entirely in the optimized design.
We also considered the economic and operational impact. The DS additive cost is approximately $15 per ton of sand, and the structural modification required only a pattern change. The overall savings from reduced scrap (from 14% to <1%) far outweigh these costs. Furthermore, the improved casting quality enhanced downstream machining and service life.
6. Conclusions
Based on our investigation and trials, we draw the following conclusions regarding the sand casting defect (shrinkage cracks) in crawler belt castings:
- Pouring temperature and steel quality have negligible influence on this sand casting defect under the studied conditions.
- The primary root cause is the high mold hardness (90 units) from the air-impact molding process, which restricts contraction and generates tensile stresses at the critical fillet.
- The casting geometry, particularly the large thermal modulus at the guide-rib fillet, exacerbates the sand casting defect. The 75.34.101C design with a 17.3 mm equivalent diameter is more prone than the 75.34.101B design (13.6 mm).
- Adding 0.5%–1.0% DS-type sand additive improves mold collapsibility and reduces the sand casting defect rate from >14% to ~1%.
- Changing the transverse reinforcement ribs to longitudinal ribs eliminates the local hot spot, and when combined with the DS additive, completely eliminates the sand casting defect.
These countermeasures have been successfully implemented in production for over two years, demonstrating a sustainable solution to this persistent sand casting defect.
Note: All data presented are from actual production records and controlled experiments. The names of specific individuals and locations have been omitted per the request.