When our company first introduced the L32/40 series large marine diesel engine, the cylinder head became one of the most challenging components we had ever produced. This component is a large ductile iron casting with a gross weight of about 600 kg and outer dimensions of 543 × 540 × 706 mm. The material specification is QT400-15, which is a ferritic ductile iron casting grade requiring excellent machinability, pressure tightness, and impact resistance. Because the cylinder head has a complex internal water jacket, valve guide bores, injector holes, and oil passages, the casting must withstand both hydraulic and pneumatic pressure tests. In our foundry, this was the first time we had dealt with such a large ductile iron casting cylinder head, and we had no mature experience to rely on. The initial trial production showed a rejection rate as high as 70%, mainly due to defects such as blowholes in the valve guide bores, gas holes in the oil gallery, shrinkage porosity on the top plane, and leakage under hydrostatic testing. In this article, I will describe the systematic analysis, simulation, and corrective actions that we developed to improve the quality of this ductile iron casting.
Original Process Design and Verification
The cylinder head is a typical thick-section ductile iron casting with heavy local masses around the valve bridges, injector boss, and bolt bosses. Because the valve guide area is structurally massive and highly susceptible to shrinkage porosity, we originally placed 45# steel rods as internal chills inside the valve guide core (air passage core). These chills were intended to accelerate solidification locally and prevent porosity. In addition, internal chills were placed in the lower water jacket area. On the lower face of the casting, external chills were placed, and insulating risers were designed on the top surface. The gating and risering system plus chill layout is described in the original design table.
| Element | Original Design | Purpose |
|---|---|---|
| Internal chills in valve guide core | 45# steel rods, uncoated, prone to rust | Local chilling to avoid shrinkage |
| Internal chills in lower water jacket | Steel bars | Avoid shrinkage in thick lower section |
| External chills on lower face | Steel plates | Directional solidification from bottom |
| Insulating risers on top | 3 risers (1#, 2#, 3#) | Feed shrinkage during solidification |
| Molding process | Alkaline phenolic resin self-hardening sand | Rigid mold for ductile iron casting |
| Core process | Triethylamine cold-box process | Complex cores with good dimensional accuracy |
| Melting | 20 t/h medium frequency induction furnace | Consistent melt quality |
| Pouring temperature | 1365–1380 °C | Fluidity and solidification control |
The molds and cores were coated with alcohol-based zirconia coating, and the chills were also painted. However, in the verification stage, the reject rate was extremely high. The main defects observed were:
- Blowholes (scab-like gas pores) inside the valve guide bores
- Gas holes on the upper surface of the oil sump
- Shrinkage porosity on the top plane
- Leakage under water pressure (hydrostatic test) around injector holes and valve guide bores
These defects are typical in ductile iron casting when gas evolution and feeding are not properly controlled. The heavy section size of this cylinder head made it even more demanding.
Root Cause Analysis of Gas-Related Defects
Gas defects in ductile iron casting are primarily caused by the combined effects of mold moisture, core gas evolution, poor venting, and surface contaminants. In our original process, the sand molds and cores were coated with alcohol-based coating, which decomposes and evolves volatiles when in contact with molten iron. The internal steel chills, especially the 45# steel rods, easily developed rust on their surfaces. Even though they were painted, the rust layer contained hydrated iron oxides. When molten iron at about 1370 °C contacts such surfaces, the rust decomposes and releases hydrogen and water vapor, which are then dissolved into the liquid metal. As the metal solidifies, these gases are rejected from the solution and form blowholes. The valve guide area is a typical location where gas pores appear, because the core is surrounded by thick metal and the gas generated at the chill surface cannot escape quickly.
The oil sump core is located at the top of the casting, far away from the ingates. The iron flow reaches that region after traveling a long distance, so the temperature is lower. In addition, the original oil sump core did not have adequate venting channels. The gas produced by the core binder and coating had no path to escape, and it remained trapped under the upper surface, creating gas holes. This is a common problem in complicated ductile iron casting core packages.
To solve these gas defects, we implemented the following measures:
- Replacement of internal chills with copper-coated steel chills. Copper-plated chills resist rust formation, thus reducing the source of hydrogen and moisture. The surface finish of the valve guide bore improved significantly.
- Double coating and baking. After applying the alcohol-based coating twice, the sand molds and cores were placed in a drying oven at 180–200 °C for 30 minutes. This removed residual moisture and low-boiling-point organics from the coating and binder.
- Enhanced venting in the mold and cores. We added more vent holes in the upper sand mold, modified the oil sump core to include built-in vent channels, and manually drilled additional vent holes in the air passage core (valve guide core) to allow gas to escape rapidly.
- Controlled time between mold assembly and pouring. After closing the mold, we poured within one hour to prevent the mold and cores from re-absorbing moisture from the atmosphere.
These changes are shown in the process comparison. The result was a dramatic reduction in gas-related rejections. In the subsequent production of 200 cylinder heads, no castings were scrapped due to blowholes, gas holes, sand inclusions, or dropped cores. This confirmed that gas evolution control is critical for sound ductile iron casting.
Shrinkage Porosity and Leakage Analysis
Shrinkage porosity is one of the most challenging defects in ductile iron casting, especially for heavy-section components like cylinder heads. The solidification mechanism of spheroidal graphite iron involves a wide freezing range between the liquidus and the end of eutectic solidification. The solidification mode is often described as “mushy” or “spongy,” because a large amount of austenite dendrites forms before the eutectic reaction. These dendrites interlock and block the liquid flow channels, making feeding very difficult. Thus, shrinkage porosity and macro-shrinkage are prone to form in isolated hot spots.
In our original design, we already used chills and insulating risers, but shrinkage defects still appeared. During pressure testing, about 60% of the scrapped castings leaked from the injector holes and valve guide holes. Destructive sectioning of these castings revealed that the leak paths were networks of internal shrinkage porosity. To understand the solidification behavior accurately, we used MAGMA simulation software to analyze the temperature field, hot spot distribution, and thermal modulus of both the casting and the risers.
Simulation of the Original Process
The simulation model included the complete gating and risering system, chills, and the QT400-15 material properties. We extracted the hot spot distribution and the thermal modulus curves along the vertical sections under each riser.
For the 1# insulating riser, the simulation showed that the hot spot was localized inside the riser and in the riser neck. No isolated hot spot remained in the casting itself. The thermal modulus of the riser interior and the riser neck was greater than the modulus of the casting body, so feeding was effective.
For the 2# insulating riser, the thermal modulus analysis showed that no isolated hot spot formed at the bottom of the riser. However, as the feeding distance increased, the thermal modulus of the casting body exhibited a peak, indicating a risk of shrinkage porosity in the casting section below the 2# riser. In other words, the feeding path was not sufficient to sustain directional solidification over the entire distance.
For the 3# insulating riser, the simulation showed that the thermal modulus of the riser interior and riser neck was actually smaller than the modulus of the casting body. This means the riser would solidify before the casting section it was supposed to feed. Consequently, an isolated hot zone formed under the 3# riser, and shrinkage porosity was highly likely in that area.
The simulated shrinkage distribution indicated two main defect regions: one under the 2# riser (region A) and one under the 3# riser (region B). These regions matched the actual leakage locations observed in the castings.
To quantify the feeding resistance, we can use the thermal modulus criterion. The modulus of a casting or riser section is defined as:
$$M = \frac{V}{A}$$
where \(V\) is the volume and \(A\) is the cooling surface area. For a riser to feed effectively, the modulus of the riser \(M_r\) must exceed the modulus of the casting section \(M_c\) by a safety factor. A common condition is:
$$M_r \ge 1.1 \cdot M_c$$
For our cylinder head, the 3# riser clearly violated this condition. The simulation data confirmed this. The modulus values for each riser are summarized below.
| Riser | Modulus of riser \(M_r\) (cm) | Modulus of casting section \(M_c\) (cm) | Feeding condition \(M_r / M_c\) | Simulated result |
|---|---|---|---|---|
| 1# insulating riser | 4.8 | 3.6 | 1.33 | Sound, no shrinkage |
| 2# insulating riser | 4.2 | 3.8 | 1.11 | Marginal, shrinkage risk in casting body |
| 3# insulating riser | 3.9 | 4.3 | 0.91 | Unsound, isolated hot spot under riser |
This table clearly explains why the original process produced leakage issues under the 3# riser and also under the 2# riser at a distance.
Improved Process: Adding Chills
Based on the simulation analysis, we decided to add additional external chills to the core package. The goal was to increase the local cooling rate and shift the hot spots from the casting body into the risers. We added two chills on the outer skin core (designated 1-1# and 1-2#) and two chills on the oil sump core (designated 2-1# and 2-2#). These chills are shown in the layout.
Chills play a dual role in ductile iron casting. First, they act as local heat sinks, increasing the solidification rate and refining the graphite structure. Second, they promote earlier graphitic expansion during solidification. When graphite nodules form, the accompanying volume expansion can feed the remaining liquid. This “self-feeding” effect is especially beneficial in thick sections, provided the mold rigidity is high enough to resist expansion. Our alkaline phenolic resin self-hardening sand mold with metal chills provided a rigid mold, so the graphitic expansion was effectively used for internal feeding.
After adding the chills, we re-ran the MAGMA simulation. The results showed that under the 2# riser, the hot spot disappeared from the casting body and moved into the riser. The thermal modulus of the casting body under the 2# riser decreased continuously toward the bottom, achieving a true directional solidification condition. The feeding condition improved dramatically.
For the 3# riser, the addition of chills reduced the hot spot size and the thermal modulus values, but the improvement was not as significant as for the 2# riser. The reason is that the 3# riser feeds a tall, narrow, thick-walled section. The riser neck diameter and the riser size were limited by the compact geometry of the cylinder head. We could not enlarge the 3# riser because of interference with the upper bolt holes and the valve arrangement. Still, the simulation showed a reduced severity of shrinkage in region B.
The new simulation of shrinkage distribution is summarized in the table below.
| Condition | Region A (under 2# riser) | Region B (under 3# riser) |
|---|---|---|
| Original process | Significant shrinkage porosity | Significant shrinkage porosity |
| New process with chills | No shrinkage porosity | Minor shrinkage porosity, reduced severity |
The simulation cross-section for the improved process showed that region A was completely sound, while region B still had some dispersed porosity, but at a much lower level.
Solidification Modeling and Feeding Efficiency
To further understand the solidification behavior, we used the Niyama criterion, which is widely used for predicting shrinkage porosity in ductile iron casting. The Niyama criterion is based on the thermal gradient \(G\) and the cooling rate \(R\) at the end of solidification. The Niyama parameter is defined as:
$$N = \frac{G}{\sqrt{R}}$$
A low value of \(N\) indicates a high risk of micro-shrinkage because the mushy zone is wide and feeding is difficult. In our simulation, we extracted the Niyama values for the critical sections before and after the process change.
| Location | Original Niyama value (°C·s\(^{1/2}\)/cm\(^2\)) | Improved Niyama value (°C·s\(^{1/2}\)/cm\(^2\)) |
|---|---|---|
| Region A under 2# riser | 0.55 | 1.12 |
| Region B under 3# riser | 0.48 | 0.72 |
| Valve bridge area | 0.62 | 1.05 |
Typically, a Niyama value above 1.0 indicates a sound casting for ductile iron casting, while values below 0.7 indicate a high risk of porosity. The improvement from 0.55 to 1.12 in region A confirmed that the chills eliminated the shrinkage problem. The improvement in region B from 0.48 to 0.72 was not enough to guarantee soundness, but it reduced the leakage rate substantially.
Another important parameter is the feeding distance \(L_f\). For ductile iron casting, the effective feeding distance of a riser depends on the section thickness and the presence of chills. A simple formula for the maximum feeding distance is:
$$L_f = L_t + L_m$$
where \(L_t\) is the thermal end zone (usually about \(2.5 \times\) section thickness) and \(L_m\) is the mid-zone (about \(4.5 \times\) section thickness for chills). With our added chills, the effective feeding distance under the 2# riser increased, allowing the riser to feed a longer length of the casting.
Production Verification of the Improved Process
After implementing the improved process, we produced 200 cylinder heads. The results were highly encouraging:
- No scrap due to blowholes, gas holes, sand inclusions, or dropped cores.
- Scraps due to top surface shrinkage porosity and leakage were reduced to 17 pieces, which is a rejection rate of only 8.5% compared to the original 70%.
- Destructive sectioning of the production castings showed that the shrinkage under the 2# riser was completely eliminated.
- The remaining leakage defects were mostly associated with the 3# riser area, as predicted by simulation.
The overall casting yield (good castings / total poured) improved dramatically. The table below summarizes the before and after performance.
| Performance indicator | Original process | Improved process |
|---|---|---|
| Total castings in verification lot | 20 (sample) | 200 |
| Gas-related rejections | 25% | 0% |
| Shrinkage/leakage rejections | 45% | 8.5% |
| Overall rejection rate | 70% | 8.5% |
| Scrap location | Valve guide, oil sump, top plane | Only under 3# riser |
The data clearly demonstrate that the systematic approach of combining defect analysis, simulation, and targeted process changes is effective for large ductile iron casting components.
Further Optimization for the 3# Riser Region
While the improved process reduced the overall rejection rate significantly, we still observed some scrap due to shrinkage in the 3# riser region. This region is a tall and narrow thick-wall section near the exhaust valve bridge. The 3# riser diameter and neck could not be increased because of the surrounding geometry, so we had to find other ways to improve feeding.
Several possibilities were considered:
- Use of exothermic sleeve on the 3# riser. An exothermic sleeve releases heat during solidification, effectively increasing the modulus without changing the physical size. The apparent modulus of an exothermic riser can be 1.3 to 1.5 times that of a conventional insulating riser.
- Increasing the chill size near the 3# riser. By placing a larger external chill on the outer wall below the 3# riser, we could increase the local cooling rate and reduce the hot spot.
- Adding an internal chill in the core near the 3# riser. This might alter the solidification sequence and promote directional solidification from the bottom toward the riser.
- Modifying the gating system to allow a hotter melt to flow into that area. The pouring temperature for ductile iron casting was already optimized at 1365–1380 °C, but perhaps a higher temperature for the last part of the pour could help.
We planned to run a second round of MAGMA simulation with an exothermic sleeve on the 3# riser and a larger chill. The target was to achieve a Niyama value above 1.0 in region B. Preliminary calculations suggested that an exothermic sleeve with a 15% modulus boost could bring the riser modulus from 3.9 cm to 4.5 cm, which would exceed the casting modulus of 4.3 cm. This should satisfy the feeding condition:
$$M_r’ = 1.15 \cdot 3.9 = 4.485 \,\text{cm} > 1.1 \cdot M_c = 4.73 \,\text{cm}$$
Actually, 4.485 is still slightly less than 4.73, so a larger boost or an additional chill would be needed. A 25% boost would give \(M_r’ = 4.875\) cm, which is greater than 4.73 cm. Thus, a high-performance exothermic sleeve might be the solution.
Alternatively, we could redesign the riser neck to have a larger cross-section while keeping the riser body same, but that might cause a hot spot at the neck, which can be cured by chills. An experimental approach with different chill sizes and a re-designed riser neck is planned.
Key Learning Points for Ductile Iron Casting Process Design
Through this extensive analysis and improvement project, I learned several important lessons that are applicable to any large ductile iron casting:
1. Gas control is primary.
In ductile iron casting, the graphite precipitation during solidification already provides a risk of gas porosity if the liquid is saturated with gas. Moisture, rust, organic binders, and coatings are all sources of hydrogen, oxygen, and nitrogen. For heavy-section castings, these gases cannot escape quickly from the thick walls. Therefore, all chills should be clean and rust-free. Copper-coated chills are highly recommended. Baking the molds and cores after coating at 180–200 °C for 30 minutes is an effective way to remove moisture. Venting must be generous, especially in cores that are far from the ingates.
2. Simulation is essential for understanding solidification.
MAGMA simulation provided crucial insights into the hot spot locations and thermal modulus relationships. Without simulation, we would not have known why the 2# and 3# risers failed. The simulated shrinkage distribution matched the actual defects well. Using the thermal modulus and Niyama criterion, we could predict the risk quantitatively and evaluate corrective actions before spending money on trial-and-error.
3. Chills can enhance self-feeding in ductile iron casting.
Chills not only remove local heat but also accelerate the formation of a solid shell. This shell increases the internal pressure during the remaining solidification, helping to feed the mushy zone. Moreover, early solidification of the outer layer means the graphitic expansion occurs under a rigid mold, reducing the need for external risers. In our new process, the added chills completely solved the shrinkage under the 2# riser, which demonstrates the power of this approach.
4. Riser modulus must exceed the required feeding modulus.
The simple modulus formula \(M = V/A\) is a powerful tool. For an insulating riser, the effective modulus is often higher than the geometric modulus because of the insulating material, but we must account for the actual heat transfer. The thermal modulus curves from simulation give a more realistic picture than simple hand calculations. In our case, the 3# riser was undersized relative to the casting section, and the simulation clearly showed this.
5. Process discipline is critical.
Keeping the time between mold closing and pouring short prevents re-absorption of moisture. Consistent coating thickness and drying cycles are essential. The foundry environment, especially in humid seasons, can affect the quality of ductile iron casting. We now have a strict protocol to ensure that all molds are poured within one hour after closing.
Conclusion
In this project, I successfully analyzed and solved the major casting defects in the L32/40 large ductile iron casting cylinder head. The original process suffered from a 70% rejection rate due to gas defects, shrinkage porosity, and leakage. By implementing copper-coated chills, enhanced drying, improved venting, and additional chills in critical areas, we reduced the rejection rate to 8.5%. The MAGMA simulation was instrumental in identifying the inadequate feeding conditions under the 2# and 3# risers. The new process completely eliminated shrinkage under the 2# riser, while the 3# riser area still requires further improvement. We are currently investigating exothermic sleeves and a redesigned riser neck to solve the last remaining shrinkage issue.
This experience has reinforced my belief that ductile iron casting is a science that requires careful control of both thermodynamics and kinetics. The combination of process simulation, thermal modulus analysis, and practical experience is the most efficient way to achieve robust production of large ductile iron castings. Our company now uses the improved process for all L32/40 cylinder heads, and the castings have performed excellently in service. We continue to refine the process for the 3# riser region, confident that we will eventually achieve zero scrap from shrinkage porosity.
In summary, the key takeaways for foundry engineers working on similar ductile iron casting components are:
- Always control gas sources: clean chills, dry molds, and good venting.
- Use simulation to predict hot spots and modulus mismatches.
- Apply chills to promote self-feeding and directional solidification.
- Ensure riser modulus is sufficiently larger than the casting modulus, including the effect of feeding distance.
- Validate with production trials and sectioning to confirm simulation results.
By sharing this analysis, I hope to help other foundries avoid similar pitfalls in producing large, pressure-tight ductile iron casting components such as cylinder heads, valve bodies, and compacted graphite iron blocks. The same principles apply to any alloy that freezes over a wide range, but they are especially critical for ductile iron casting due to its complex solidification and graphitic expansion behavior.
Our journey with the L32/40 cylinder head is a perfect example of how a structured method, combining defect analysis, numerical simulation, and foundry practice, can turn a problematic ductile iron casting into a reliable, high-quality product. I am confident that the remaining issue at the 3# riser will be resolved soon, and we will continue to apply these techniques to other new ductile iron casting projects in the future.