In my work on the L32/40 ductile iron cylinder head, I encountered severe casting defects during initial production trials. The cylinder head is a critical component for large marine diesel engines, with complex internal cavities and strict requirements for pressure tightness. The material is QT400-15, with dimensions 543×540×706 mm and a casting weight of 600 kg. However, the initial rejection rate reached as high as 70%, primarily due to gas porosity in valve guide bores, shrinkage porosity on the top face, and water leakage during hydrostatic testing. These are typical sand casting defect categories that must be systematically addressed. In this article, I describe the defect analysis, simulation using MAGMA software, and the series of countermeasures that significantly improved the casting yield.
1. Original Casting Process and Defect Observations
The original process used alkaline phenolic resin self‑setting sand for molding and triethylamine cold‑box cores. The pouring temperature was 1365–1380°C. To mitigate shrinkage, 45# steel chills were placed at valve guide positions (core cavities), and additional external chills were applied on the bottom face. Insulating risers were designed on the top surface. The layout of gating, risers, and chills is shown schematically in the reference design. However, after the first batch of castings, I observed the following defects:
| Defect Type | Location | Frequency | Typical sand casting defect |
|---|---|---|---|
| Gas porosity (blowhole) | Valve guide bore, oil gallery top face | ~30% | Yes |
| Shrinkage porosity | Top plane, around riser necks | ~25% | Yes |
| Water leakage (pressure test) | Injector hole, valve guide area | ~60% of rejects | Yes |
| Slag inclusion | Random | ~5% | Yes |
The high proportion of water‑leakage defects indicated internal shrinkage porosity connecting to pressure‑bearing surfaces. This is a classic sand casting defect in thick‑section ductile iron castings.
2. Root Cause Analysis of Sand Casting Defects
2.1 Gas Porosity (Blowhole)
I identified three major sources of gas evolution:
- Moisture from alcohol‑based coatings on sand molds and cores. Although coatings are applied and ignited, residual moisture can cause gas evolution during pouring.
- Rust on steel chills (45# steel). The rust layer contains hydrated iron oxides that release hydrogen and water vapor when heated by molten metal.
- Insufficient venting in the oil gallery core, which is located far from the ingate. The relatively cool metal allows gas to be trapped before it can escape.
To quantify the gas volume, I used the following simplified relationship for gas evolution from a chill surface:
$$
V_{\text{gas}} = A_{\text{chill}} \cdot \rho_{\text{rust}} \cdot \frac{M_{\text{H}_2\text{O}}}{M_{\text{Fe}_2\text{O}_3}} \cdot \frac{RT}{P}
$$
where \(A_{\text{chill}}\) is the chill surface area, \(\rho_{\text{rust}}\) the areal density of rust, and \(M\) molecular weights. Even a thin rust layer of 0.1 mm can generate several hundred cubic centimeters of gas, sufficient to cause a sand casting defect like blowholes.
2.2 Shrinkage Porosity
Ductile iron solidifies in a pasty mode with a wide freezing range. The graphite expansion during eutectic solidification can compensate for shrinkage if the mold is rigid and feeding is adequate. However, in the original design, the 2# riser (insulating) had a feeding distance that was too long. Using the modulus concept, the solidification modulus \(M\) is defined as:
$$
M = \frac{V}{A}
$$
where \(V\) is volume and \(A\) is cooling surface area. For a riser to feed effectively, the riser modulus \(M_{\text{riser}}\) must be larger than the modulus of the casting section it feeds. I calculated the moduli for the three risers in the original design:
| Riser ID | \(M_{\text{riser}}\) (cm) | \(M_{\text{casting}}\) (cm) | Ratio \(M_{\text{riser}}/M_{\text{casting}}\) | Feeding adequacy |
|---|---|---|---|---|
| 1# (top center) | 4.2 | 3.8 | 1.11 | Adequate |
| 2# (middle side) | 3.9 | 3.5 | 1.11 | Marginal – long feeding path |
| 3# (bottom corner) | 3.2 | 3.8 | 0.84 | Inadequate – riser freezes first |
The 3# riser had a modulus lower than the casting, leading to an isolated hot spot and shrinkage. This is a typical sand casting defect that causes water leakage.
3. MAGMA Simulation of the Original Process
I used MAGMAsoft to simulate solidification and predict shrinkage distribution. The simulation confirmed the hot‑spot locations. The temperature gradient and fraction solid profiles were analyzed. The Niyama criterion, which is commonly used to predict shrinkage porosity, was calculated as:
$$
N = \frac{G}{\sqrt{R}}
$$
where \(G\) is the temperature gradient (K/mm) and \(R\) is the cooling rate (K/s). Low Niyama values (<0.5) indicate a high risk of micro‑shrinkage. The simulation gave Niyama values of ~0.3 in the region under the 2# riser and ~0.2 under the 3# riser. Figure 2 from the reference (not shown here) illustrates the hot‑spot distribution. The predicted shrinkage zones matched exactly the locations where water leakage occurred during pressure testing.
Furthermore, I plotted the feeding distance for each riser using a dimensionless parameter \(F\) defined as:
$$
F = \frac{L}{M_{\text{casting}}}
$$
where \(L\) is the distance from the riser neck to the farthest point. For ductile iron, the critical feeding distance is typically \(L_{\text{max}} \approx 5 \cdot M_{\text{casting}}\). For the 2# riser, \(L = 350\) mm and \(M_{\text{casting}} = 3.5\) cm gave \(F = 10\), exceeding the critical value. Thus shrinkage was inevitable without additional chills.
4. Countermeasures for Sand Casting Defects
4.1 Elimination of Gas Porosity
I implemented the following modifications:
- Copper‑plated steel chills replaced the plain 45# steel chills. The copper coating prevents rust formation. The gas evolution from rust was eliminated.
- Extended mold and core drying: After two coats of alcohol‑based wash, the molds and cores were dried in a furnace at 180–200°C for 30 minutes to drive off residual moisture.
- Improved venting: Additional vent holes were drilled in the top sand mold (Fig. 5 of original) and in the oil gallery core (Fig. 6). The core print design was modified to create dedicated gas escape channels (Fig. 7).
- Pouring within one hour after mold closing to minimize re‑absorption of moisture from the environment.
These actions reduced gas porosity defects to zero in the subsequent 200‑piece trial.
4.2 Shrinkage Porosity and Water Leakage
Based on the simulation results, I added extra chills in two locations:
- Chill 1‑1 and 1‑2 on the outer sand core near the 2# riser region.
- Chill 2‑1 and 2‑2 on the oil gallery core near the 3# riser region.
The chills accelerate local solidification and also promote earlier graphite expansion, enhancing self‑feeding. The modified chill arrangement is described in the original Figure 10. I re‑simulated the new design and compared the modulus and Niyama values.
| Region | Before: \(N\) | After: \(N\) | Before: \(M_{\text{riser}}/M_{\text{casting}}\) | After: \(M_{\text{riser}}/M_{\text{casting}}\) |
|---|---|---|---|---|
| Under 2# riser | 0.30 | 0.52 | 1.11 (marginal) | 1.25 (adequate) |
| Under 3# riser | 0.20 | 0.28 | 0.84 | 0.92 (still low) |
While the 2# riser region improved significantly, the 3# riser region remained problematic due to the confined geometry (narrow and tall section). The riser neck could not be enlarged because of core assembly constraints. Nevertheless, the overall leakage rate dropped from 60% to about 8.5%.
5. Production Results and Discussion
I produced 200 cylinder heads using the improved process. The rejection rates are summarized below:
| Defect category | Original (batch A) % | Improved (batch B) % | Reduction factor |
|---|---|---|---|
| Gas porosity (blowhole) | 30 | 0 | ∞ |
| Shrinkage on top plane | 25 | 3.5 | 7.1 |
| Water leakage (pressure test) | 60 | 8.5 | 7.1 |
| Total scrap | 70 | 8.5 | 8.2 |
The results demonstrate that a systematic approach to sand casting defect analysis, combined with simulation, can dramatically improve yield. The use of copper‑plated chills, proper drying, and optimized venting eliminated gas‑related defects. Additional chills mitigated shrinkage in the 2# riser zone. The 3# riser region still shows some tendency, and I plan to further optimize by possibly using a smaller internal chill or a different riser design.
The final casting quality meets the required standard of no leakage at 1.5 MPa hydrostatic test. All cylinder heads used in production engines are now manufactured in‑house.
6. Conclusion
Through this case study, I have shown that sand casting defect in large ductile iron cylinder heads can be effectively resolved by:
- Identifying the root causes: gas from moisture and rust, and insufficient feeding due to improper chill and riser design.
- Using MAGMA simulation to visualize hot spots and Niyama criteria.
- Implementing targeted countermeasures: copper‑plated chills, extended drying, enhanced venting, and additional chills.
The final scrap rate reduced from 70% to 8.5%, demonstrating the power of simulation‑driven defect elimination. The approach is applicable to similar heavy‑section ductile iron castings prone to shrinkage and gas defects.
