In our foundry, the production of gray iron bed castings for machine tools had historically been stable using clay sand molds and dry cores. However, after outsourcing to an external foundry that employed sodium silicate sand (water glass) molds, we encountered severe sand casting defects, specifically subsurface blowholes and oxide slag inclusions in the V‑shaped guide rail surfaces. These defects led to rejection rates exceeding 40% during final machining, causing significant economic losses. This article details our systematic investigation into the root causes of these sand casting defects and the effective preventive measures we implemented, emphasizing the critical role of residual moisture in sodium silicate sand molds.
1. Background and Defect Characteristics
The casting under study is a 145 kg gray iron bed (HT200) for a 2M9120A multi‑purpose grinder. The original process used clay sand, dry molds and cores, with a horizontal split pattern and top risers. After outsourcing, the external foundry switched to sodium silicate sand bonded with CO₂ hardening, while retaining the same gating system layout. During rough machining and artificial aging, no issues appeared. However, during finish machining of the V‑shaped guide rails, numerous subsurface pores and dark oxide slag spots were discovered. These sand casting defects were concentrated in the upper sections of the rail faces, often appearing as small, round or elongated cavities filled with oxide scales or glassy material.
The characteristic features of these sand casting defects included:
- Subsurface location – only revealed after 2–3 mm of machining.
- Irregular shapes with smooth inner walls, sometimes interconnected.
- Oxide films and slag residues adhering to the pore walls.
- Predominant occurrence in heavy sections with slow cooling rates.
Table 1 summarizes the observed defect types and their typical locations.
| Defect Type | Appearance | Location | Frequency |
|---|---|---|---|
| Subsurface blowhole | Spherical or elongated, 0.5–3 mm | V‑rail top surface | ~35% |
| Oxide slag inclusion | Irregular dark patches, 1–5 mm | Under surface layer | ~25% |
| Mixed gas‑slag defect | Pores with oxide filling | Near gate area | ~10% |
| Shrinkage porosity | Dendritic cavities | Heavy section center | ~5% |
2. Root Cause Analysis of Sand Casting Defects
We formed a joint quality team with the external foundry and performed a thorough investigation. The key factors contributing to these sand casting defects were identified as follows.
2.1 Residual Moisture in Sodium Silicate Sand Molds
Sodium silicate sand after CO₂ hardening retains a significant amount of chemically bound and free water. The residual moisture content can be as high as 3–5 wt%. During pouring, the intense heat decomposes the water into hydrogen and oxygen, which can be absorbed by the molten iron. The solubility of hydrogen in liquid iron follows Sieverts’ law:
$$ \frac{1}{2} H_2 (g) \rightarrow [H] \quad ; \quad K = \frac{[H]}{\sqrt{p_{H_2}}} $$
At pouring temperatures around 1400 °C, the equilibrium hydrogen concentration in molten iron can easily exceed 4 ppm. When the metal solidifies, the sudden drop in solubility forces hydrogen to precipitate, forming blowholes. Moreover, the oxygen from water decomposition reacts with carbon in the iron to form CO gas:
$$ C + \frac{1}{2} O_2 \rightarrow CO \uparrow $$
These gas‑forming reactions are especially severe when the mold’s permeability is low. The external foundry used fine sand and low‑modulus sodium silicate, resulting in a dense mold face with poor gas permeability. Table 2 compares the mold properties between the original clay‑sand process and the problematic sodium silicate process.
| Property | Clay Sand (Dry) | Na₂SiO₃ Sand (CO₂) |
|---|---|---|
| Moisture content (wt%) | 0.5–1.0 | 3.5–5.0 |
| Permeability (AFS) | 120–160 | 60–90 |
| Compressive strength (MPa) | 0.3–0.5 | 0.8–1.2 |
| Surface stability | Good | Moderate |
2.2 Gating System Design and Venting
The original gating system (as used in clay sand) had a single sprue, a cross runner, and five small vent risers on the top mold. This layout was adequate for dry clay molds but insufficient for sodium silicate molds because the gases generated from moisture decomposition are much larger in volume. The pressure of gas evolved can be estimated by the ideal gas law:
$$ PV = nRT $$
For a casting with a mold cavity volume of ~20 L, if 100 g of water is decomposed at 1400 °C, the volume of steam generated at atmospheric pressure would be:
$$ V = \frac{nRT}{P} = \frac{(100/18) \times 8.314 \times 1673}{101325} \approx 7.6\ \text{m}^3 $$
Such a huge volume of gas must be vented rapidly; otherwise, it creates back‑pressure that forces gas into the solidifying metal. The original five vent risers provided only about 15 cm² of total exit area, which was grossly inadequate.
2.3 Pouring Temperature and Speed
The external foundry used medium‑frequency induction furnaces, capable of superheating the melt above 1500 °C. They poured at temperatures as high as 1420–1450 °C. While high temperature improves fluidity, it also increases hydrogen solubility (endothermic reaction) and promotes oxidation. The oxidation reaction:
$$ 2Fe + O_2 \rightarrow 2FeO $$
Iron oxide (FeO) has a low melting point (~1370 °C) and can form immiscible slag droplets that become entrapped. Furthermore, pouring speed was not controlled; the operators often poured too fast, causing turbulent flow that engulfed air and slag.
2.4 Metallurgical Factors
The return scrap used in the charge contained nodular iron risers from previous heats. The residual magnesium and rare earth elements from the nodular iron promoted the formation of magnesium oxide and rare earth oxides, which are high‑melting‑point inclusions. These act as nucleation sites for gas pores. Additionally, the high sulfur content (0.08–0.12 wt%) in the iron increased the tendency for hydrogen pickup, because sulfur reduces the surface tension of the melt, making it easier for gas bubbles to nucleate:
$$ \gamma = 1.8 – 0.28\ [\%S] \quad (\text{N/m}) $$
Table 3 lists the chemical composition ranges before and after corrective actions.
| Element | Before (Defective) | After (Corrected) |
|---|---|---|
| C | 3.2–3.4 | 3.1–3.3 |
| Si | 1.8–2.0 | 1.9–2.1 |
| Mn | 0.6–0.8 | 0.7–0.9 |
| P | ≤0.15 | ≤0.12 |
| S | 0.08–0.12 | 0.06–0.10 |
| Mg (residual) | 0.01–0.03 | <0.005 |
| RE (residual) | 0.005–0.015 | <0.002 |
3. Preventive Measures
Based on the root cause analysis, we implemented a series of corrective actions. These measures can be grouped into three categories: (A) control of mold moisture and permeability, (B) optimization of gating and venting, and (C) improvement of melting and pouring practices.
3.1 Control of Mold Moisture and Permeability
We switched to high‑modulus sodium silicate (M = 2.8–3.0) and coarser silica sand (AFS fineness 40–55 instead of 70–90). The binder content was reduced from 6% to 4.5%. After CO₂ hardening, the molds and cores were air‑dried for at least 4 hours, then coated with a fast‑drying alcohol‑based zirconia wash. Finally, the cores were baked in an oven following the drying schedule shown below (Figure 3 in the original, represented here as a table).
| Stage | Temperature (°C) | Time (min) | Purpose |
|---|---|---|---|
| Pre‑heat | 100–120 | 30 | Even heating |
| Drying | 180–200 | 60 | Remove capillary water |
| Dehydration | 250–280 | 90 | Remove chemically bound water |
| Cooling | ≤60 | ~120 | Prevent re‑absorption |
The resulting mold properties improved dramatically, as shown in Table 5.
| Property | Before Correction | After Correction |
|---|---|---|
| Residual moisture (%) | 3.5–5.0 | <1.0 |
| Permeability (AFS) | 60–90 | 130–160 |
| Surface hardness (B scale) | 85–90 | 75–80 |
| Hot strength (MPa at 1000°C) | 0.4–0.6 | 0.8–1.0 |
3.2 Improved Gating and Venting System
We redesigned the gating system as follows:
- Increased sprue height from 200 mm to 350 mm to increase metallostatic pressure for better feeding and gas removal.
- Added a large riser (φ100 mm, 150 mm high) between the sprue and the mold cavity to act as a slag trap and gas collector.
- Increased the number of vent risers from 5 to 8, each with a diameter of 20 mm, providing a total vent area of 25 cm².
- Inclined the mold during pouring: the side containing the sprue was raised by 150 mm, so that the metal fills the cavity from the lower end and gases escape upward toward the vents.
The effective feeding distance of the riser can be calculated using the shape factor method. For a rectangular plate, the feeding distance L is:
$$ L = A \times \sqrt{\frac{V}{W}} $$
where A is a constant (typically 15–20 for gray iron), V is the riser volume, and W is the casting weight. With the new riser, V ≈ 1.2 L, W = 145 kg, giving L ≈ 480 mm, which covers the entire length of the guide rail (approx. 900 mm when using two risers).
Table 6 compares the original and modified gating parameters.
| Parameter | Original | Modified |
|---|---|---|
| Sprue height (mm) | 200 | 350 |
| Sprue diameter (mm) | 25 | 30 |
| Cross‑runner area (mm²) | 400 | 600 |
| Number of ingates | 2 | 4 |
| Number of vent risers | 5 | 8 |
| Additional slag riser | None | 1 (φ100) |
| Pouring tilt angle (°) | 0 | 5 |
3.3 Melting and Pouring Control
We eliminated the use of nodular iron returns and strictly segregated all scrap. The charge consisted of 60% pig iron, 30% clean steel scrap, and 10% gray iron returns. The melting temperature was limited to 1500 °C maximum, and the pouring temperature was controlled between 1360 °C and 1390 °C. A slag dam was installed at the pouring basin, and operators were trained to follow a “slow‑fast‑slow” pouring sequence: initial slow pour to fill the sprue gently, a fast fill of the cavity (within 8–10 seconds), and a final slow pour to top off the risers. Inoculation with 0.3% ferrosilicon was performed in the ladle to refine graphite and reduce chill tendency.
The dissolved hydrogen content in the melt was measured using a reduced‑pressure test. The target was <4 ppm. The relationship between hydrogen content and pore formation can be expressed by the critical supersaturation ratio:
$$ S = \frac{[H]_{\text{melt}}}{[H]_{\text{saturation}}} $$
When S > 1.2, nucleation of hydrogen bubbles becomes likely. By controlling the melt temperature and composition, we kept S below 1.1.
4. Results and Discussion
After implementing these measures in November 2007, we produced a batch of 63 bed castings. Only 3 castings exhibited minor sand casting defects on the V‑rails, resulting in a rejection rate of less than 5%. This was a dramatic improvement from the initial 40% and even from the 25% after the first round of partial corrections.
Table 7 summarizes the defect rates throughout the evolution of the process.
| Stage | Number of Castings | Defective Castings | Defect Rate (%) |
|---|---|---|---|
| Initial (external foundry) | ~30 | 12+ | 40 |
| After partial melt control | 12 | 3 | 25 |
| After full corrective actions | 63 | 3 | 4.8 |
The success confirms that residual moisture in sodium silicate sand is the primary driver of these sand casting defects. Even after CO₂ hardening, water remains chemically bonded in the sodium silicate structure. The drying schedule we adopted (Table 4) effectively removes both free and bound moisture. The baking curve follows the principle that water desorption is a diffusion‑controlled process, which can be modeled by Fick’s law:
$$ \frac{\partial C}{\partial t} = D \frac{\partial^2 C}{\partial x^2} $$
where C is moisture concentration, D is diffusion coefficient (temperature‑dependent). The holding time at 250–280 °C ensures that the core center reaches the required dryness.
The combination of enhanced venting and the tilted pour not only improves gas escape but also promotes directional solidification. Solidification of the casting can be modeled with the thermal modulus approach. The modulus M = V/A (volume/surface area) of the guide rail section is about 2.5 cm. The riser modulus was designed to be 1.2 times larger, i.e., 3.0 cm, ensuring that the riser solidifies last and provides adequate feed.
We also observed that the oxide slag defects diminished because the slag riser trapped the dross, and the slow‑fast‑slow pouring minimized re‑oxidation. The turbulent kinetic energy in the melt can be estimated by:
$$ E_k = \frac{1}{2} \rho v^2 $$
By reducing the pouring rate from 15 kg/s to 10 kg/s during the fast stage, the kinetic energy decreased by about 55%, leading to less surface entrainment.
5. Conclusion
From this investigation, we draw the following conclusions regarding the prevention of sand casting defects in gray iron produced with sodium silicate sand:
- Residual moisture is the most critical factor – It directly causes hydrogen‑ and CO‑related blowholes. Using high‑modulus binder, coarser sand, and thorough drying of cores reduces moisture to below 1% and eliminates this primary source of sand casting defect.
- Proper gating and venting design is essential – Increasing sprue height, adding slag risers, augmenting vent area, and tilting the mold during pouring significantly improve gas removal and reduce turbulence. The vent area should be calculated based on the expected gas volume from moisture decomposition, which can be estimated from the ideal gas law.
- Metallurgical cleanliness matters – Contamination from nodular iron returns introduces magnesium and rare earth oxides that nucleate pores. Strict segregation of scrap and controlled pouring temperatures (1360–1390 °C) minimize oxide formation.
- Process discipline – Training operators to follow a correct pouring sequence and to maintain consistent melt treatment is indispensable for achieving low defect rates.
The systematic approach we applied – identifying the root causes, quantifying each factor with simple engineering formulas, and implementing targeted corrections – turned a 40% sand casting defect rate into less than 5%. This case demonstrates that even with a change in molding material, the fundamental principles of gas evolution, metal flow, and solidification can be leveraged to produce defect‑free castings.
We continue to monitor the process and have since applied the same principles to other gray iron castings using sodium silicate sand, with similarly successful outcomes. The key is to never underestimate the role of moisture in sand casting defect formation.