In my years of experience with sand casting, I encountered a particularly challenging case involving a pump shell casting made of GG30 (equivalent to EN-GJL-250), weighing 28.8 kg. The production line used a German KW molding machine with flask dimensions of 800 mm × 600 mm × 300 mm/300 mm. The cores were made of resin-coated sand, and the pattern plate was arranged for four castings per mold. Melting was carried out in a 3 t/h medium-frequency induction furnace. Despite the advanced equipment, several sand casting defects emerged, threatening the quality and yield of the product. This article details my systematic approach to identifying, analyzing, and eliminating these sand casting defects, with a focus on quantitative methods and process optimization.
The three predominant sand casting defects were: shrinkage porosity (micro-porosity), core shift (dimensional deviation), and blowhole (gas porosity). Each defect had distinct causes and required tailored countermeasures. Before presenting the solutions, I will first describe the defects and their root causes in detail.
1. Shrinkage Porosity
Shrinkage porosity appeared in a localized region approximately 30 mm to 25 mm below the flange face, forming a 5 mm × 6 mm × 7 mm cavity. This area corresponded to the threaded hole location for assembling the pump shell and back cover. The defect was identified via ultrasonic testing and sectioning. The primary cause was the lack of an adequate feeding channel for the final solidification zone. In gray cast iron, graphite precipitation during eutectic solidification generates volumetric expansion that can compensate for liquid shrinkage. However, the original carbon content (3.0%–3.1%) and inoculation level (0.4%) were insufficient to fully exploit this expansion.
The graphite expansion effect can be described by the volume change during the eutectic reaction:
$$ \Delta V_{\text{graphite}} = V_{\text{graphite}} – V_{\text{liquid}} $$
where \(V_{\text{graphite}}\) is the volume of graphite precipitated, and \(V_{\text{liquid}}\) is the volume of liquid iron consumed. For gray iron, the specific volume of graphite is approximately 0.42 cm³/g, while liquid iron has a specific volume of around 0.13 cm³/g. Thus, each gram of graphite precipitated yields an expansion of about 0.29 cm³. By increasing the carbon equivalent (CE), more graphite precipitates, reducing the net shrinkage. The carbon equivalent is calculated as:
$$ \text{CE} = \text{C} + \frac{1}{3}(\text{Si} + \text{P}) $$
In this case, we raised the target carbon content from 3.0%–3.1% to 3.1%–3.2%, and increased the total inoculant addition from 0.4% to 0.5%. This adjustment shifted the solidification path toward a more graphitic eutectic, enhancing the self-feeding capability. The resulting change in carbon equivalent is shown in Table 1.
| Element | Original (wt%) | Modified (wt%) |
|---|---|---|
| C | 3.00–3.10 | 3.10–3.20 |
| Si | 1.80–2.00 | 1.80–2.00 (unchanged) |
| Mn | 0.60–0.80 | 0.60–0.80 (unchanged) |
| P | ≤0.12 | ≤0.12 |
| S | 0.08–0.12 | 0.08–0.12 |
| Cu | 0.40–0.60 | 0.50–0.70 |
| Sn | 0.05–0.08 | 0.08–0.10 |
| Inoculant (FeSi) | 0.4% (ladle) | 0.5% (ladle + stream) |
The increase in carbon equivalent from approximately 3.6–3.7 to 3.7–3.8 promoted a more pronounced graphite expansion. To retain this expansion effect, we also increased the mold hardness from 85±5 to 90±5 (measured on a standard hardness tester). Higher mold rigidity prevents the mold wall from yielding under expansion pressure, maximizing the internal pressure that helps feed the shrinkage zones.
Additionally, to maintain the required tensile strength (≥250 MPa), we slightly raised the copper and tin contents to stabilize pearlite. The relationship between carbon equivalent and tensile strength for gray iron can be approximated by:
$$ R_m (\text{MPa}) = 1000 – 800 \cdot \text{CE} $$
For the original CE of 3.65, \(R_m \approx 1000 – 800 \times 3.65 = 1000 – 2920 = -1920\) (not valid). This formula is only empirical for certain ranges. In practice, strength is more directly related to carbon content and cooling rate. By controlling the copper (0.5%–0.7%) and tin (0.08%–0.10%), we achieved a fine pearlitic matrix with tensile strength consistently above 260 MPa, even with the higher carbon content.
2. Core Shift (Dimensional Deviation)
The second sand casting defect was core shift, detected during dimensional inspection. The wall thickness at the tail end of the pump shell (near the Ø55 mm hole) showed a significant deviation: the lower wall was 2.2 mm–2.5 mm thicker than the upper wall, exceeding the tolerance of ±0.8 mm. This indicated that the core had floated upward during pouring. Core shift is a typical sand casting defect caused by buoyancy forces exceeding the core print restraint, inadequate core support, or uneven sand compaction.
The buoyancy force acting on a core can be calculated as:
$$ F_b = \rho_{\text{liquid}} \cdot g \cdot V_{\text{core}} $$
where \(\rho_{\text{liquid}}\) is the density of molten iron (approximately 7000 kg/m³), \(g\) is gravity (9.81 m/s²), and \(V_{\text{core}}\) is the volume of the core submerged in the liquid. For the pump shell core, the estimated volume was 0.0008 m³, resulting in \(F_b \approx 7000 \times 9.81 \times 0.0008 \approx 55 \text{ N}\). While this force is modest, the core print area and the sand strength must provide sufficient frictional resistance.
To counteract the core shift, I redesigned the core print in the pattern. Instead of centering the core print at the nominal position, I offset the Ø55 mm core print seat in the bottom mold by 1.2 mm downward. This pre-compensation accounted for the expected upward movement of the core during pouring. The offset was determined by iterative trials: measurements showed that the core shifted upward by an average of 1.1 mm; by adding a 0.1 mm safety margin, the 1.2 mm offset was selected. The resulting wall thickness variation was reduced to less than 0.7 mm, well within specification.
| Parameter | Before Correction | After Correction |
|---|---|---|
| Lower wall thickness (mm) | 10.2 ± 0.5 | 8.5 ± 0.3 |
| Upper wall thickness (mm) | 8.0 ± 0.4 | 8.2 ± 0.3 |
| Maximum deviation (mm) | 2.2 – 2.5 | 0.3 – 0.7 |
| Dimensional acceptance | Fail | Pass |
The pre-offset method effectively compensated for the buoyancy-induced core shift. It is important to note that the offset amount must be validated for each geometry; I used a series of trial castings to optimize the value. This approach eliminated the core shift sand casting defect without altering the core weight or the filling pattern.
3. Blowhole (Gas Porosity)
The most severe sand casting defect in terms of scrap rate was blowhole formation at a specific location (originally referred to as point D). The blowholes were exposed on the casting surface, typically one per casting, with diameters ranging from 6 mm to 15 mm and depths of 3 mm to 6 mm. The defect scrapped 45% of the production lot. Blowholes are caused by trapped gas—either from core binder decomposition, mold moisture, or entrapped air—that cannot escape before the metal solidifies.
In this case, the gas source was likely the resin-coated sand core. During pouring, the binder (phenolic resin) decomposes, generating gases such as CO, CO₂, H₂, and hydrocarbons. Proper venting is critical. The gas volume generated can be estimated by:
$$ V_{\text{gas}} = m_{\text{binder}} \cdot Y_{\text{gas}} $$
where \(m_{\text{binder}}\) is the mass of binder in the core (about 2.5% of core weight), and \(Y_{\text{gas}}\) is the gas yield (typically 150–200 cm³ per gram of binder). For a core weighing 1.2 kg, the binder mass is ~30 g, producing 4500–6000 cm³ of gas at high temperature. If this gas cannot escape, it forms blowholes.
My initial idea was to install a vent pin directly through the mold cavity at the blowhole location. However, this posed two problems: (1) after removing the vent pin, the resulting cavity (if not properly filled) would leave a depression (missing metal) on the casting surface; (2) the root of the vent pin could itself become a gas trap and nucleate a blowhole. Therefore, I chose to place the vent system on the core print, connected to the cavity via a 1 mm thick vent strip. The vent pin (or a small vent hole) was drilled into the core print seat on the pattern, and a shallow groove (1 mm deep × 3 mm wide) was cut from the casting cavity to the vent pin. This provided a direct path for gas to escape into the core print and then to the mold atmosphere, without leaving any features on the final casting.
The modified venting design is illustrated in the following image. The figure shows a typical sand casting defect scenario where gas porosity occurs; the venting solution is similar in principle. (Please refer to the embedded image below.)
Table 3 summarizes the venting parameters before and after the modification.
| Parameter | Original Design | Modified Design |
|---|---|---|
| Vent location | None (no dedicated vent) | Core print seat |
| Vent connection to cavity | — | 1 mm thick × 3 mm wide vent strip |
| Number of vents per mold | 0 | 4 (one per cavity) |
| Blowhole scrap rate | 45% | <1% |
The vent strip was 1 mm thick and 3 mm wide, running from the cavity surface to the core print. This thickness was chosen to be small enough that the strip would be easily filled by liquid iron without causing a cold shut, yet large enough to allow gas passage. The vent pin (0.5 mm diameter) was embedded in the core print seat of the mold and extended to the back of the pattern plate, where the gas was released to the atmosphere. This system eliminated the blowhole sand casting defect effectively.
4. Overall Results and Discussion
After implementing all three corrective measures, I conducted a series of trials: first a few test castings, then a pilot batch of 200 pieces, and finally full-scale production. The results are summarized in Table 4.
| Defect Type | Before Improvement | After Improvement |
|---|---|---|
| Shrinkage porosity | 5 mm × 6 mm × 7 mm cavity (detected by UT), requiring rejection | Reduced to <2 mm × 2 mm × 2 mm (dispersed micro-porosity), completely removed during threading operation |
| Core shift (wall thickness deviation) | 2.2 mm – 2.5 mm (fail) | ≤0.7 mm (pass) |
| Blowhole (exposed) | 45% scrap rate | <1% scrap rate |
| Overall casting yield | ~50% (estimated) | >95% |
The shrinkage porosity, although not completely eliminated, was reduced to a level that did not affect the functional integrity. The micro-porosity zone was smaller than the machining allowance for the threaded holes; thus, after drilling and tapping, no remnant porosity remained on the finished surfaces. This was confirmed by sectioning and microscopic examination.
The success of these modifications demonstrates that a systematic approach to sand casting defects—understanding the physics of each defect and applying targeted process changes—can dramatically improve quality. The key factors were:
- For shrinkage: increasing carbon content and inoculation promotes graphite expansion, and higher mold hardness retains that expansion.
- For core shift: offsetting the core print compensates for buoyancy forces.
- For blowholes: providing a direct vent path from the cavity to the core print without leaving potential defect nuclei on the casting surface.
Quantitative modeling of these phenomena can further refine the process. For instance, the total volumetric change during solidification can be expressed as:
$$ \Delta V_{\text{total}} = V_{\text{solidification contraction}} – V_{\text{graphite expansion}} + V_{\text{gas porosity}} $$
By controlling the balance, one can minimize the net contraction. The critical feeding distance can also be estimated using Chvorinov’s rule:
$$ t_s = B \left( \frac{V}{A} \right)^n $$
where \(t_s\) is the solidification time, \(V/A\) is the volume-to-surface area ratio (modulus), and \(B\) and \(n\) are constants. In this pump shell, the region with shrinkage porosity had a lower modulus than the adjacent heavy sections, making it a hot spot. By adjusting the chemical composition to increase the expansion, the need for external feeding was reduced.
Finally, I would emphasize that preventing sand casting defects requires both theoretical knowledge and practical experimentation. The combination of chemical adjustments, pattern modifications, and venting improvements eliminated the three major defects, raising the yield from approximately 50% to over 95%. This case study can serve as a reference for similar castings prone to these sand casting defects.
5. Conclusion
In conclusion, the pump shell casting suffered from three critical sand casting defects—shrinkage porosity, core shift, and blowholes. Through systematic analysis and targeted modifications, each defect was successfully mitigated:
- Shrinkage porosity was reduced by increasing carbon and inoculant levels, and by raising mold hardness.
- Core shift was corrected by offsetting the core print 1.2 mm downward in the pattern.
- Blowholes were eliminated by adding vent strips from the cavity to the core print, avoiding direct vent pins on the casting.
These changes were validated through trials and mass production, resulting in a scrap rate reduction from 45% to below 1% for blowholes, and acceptable levels for the other defects. The successful resolution of these sand casting defects highlights the importance of a data-driven, engineering approach in foundry practice.