In the course of developing a large marine diesel engine series, our company encountered significant challenges in producing a ductile iron cylinder head. The component, designated as L32/40, is a critical part of the engine and is subject to stringent quality requirements including water pressure and air pressure tests. The casting material is QT400-15, with external dimensions of 543 mm × 540 mm × 706 mm and a rough weight of 600 kg. This was our first attempt at such a large ductile iron casting, and no mature experience was available for reference. The initial production trials resulted in a very high rejection rate of up to 70%, with defects such as gas holes in the valve guide bores, pinholes on the oil gallery top surface, shrinkage porosity on the top plane, and leakage during hydrostatic testing. This article describes the systematic analysis, simulation-assisted process optimization, and the successful implementation of corrective measures that substantially improved the yield of these ductile iron castings.
Ductile iron castings, also known as spheroidal graphite iron castings, are widely used for demanding structural applications because of their excellent combination of strength, ductility, and machinability. However, the solidification behavior of ductile iron presents unique challenges, especially for thick-walled components. The mushy zone is wide, and the volumetric expansion associated with graphite precipitation can either compensate for shrinkage or, if uncontrolled, lead to internal porosity. In this paper, we focus on the practical aspects of eliminating major casting defects in a complex geometry, emphasizing the role of cooling chills, exothermic sleeves, venting, and simulation tools such as MAGMA.
Original Casting Process and Initial Validation
The cylinder head geometry is shown in the figure above. The internal water jacket, oil galleries, valve guide bores, and injector pockets create a complex array of cores and intricate sections. In the original design, the process employed the following key elements:
- Molding: Alkaline phenolic resin self-setting sand.
- Core making: Triethylamine cold-box process.
- Melting: 20 t/h medium frequency induction furnaces.
- Pouring temperature: 1365–1380 °C.
- Cooling chills: 45# steel rods placed inside the valve guide cores (air intake & exhaust passages) to act as internal chills for local rapid cooling.
- Additional internal chills were set in the lower water jacket.
- External chills were placed on the lower parting plane.
- Insulating risers (exothermic sleeves) were placed on the top surface.
Figure 2 in the original article (not reproduced here) showed the layout of the gating/riser system and chills. The risers were located at three positions: 1# riser, 2# riser, and 3# riser. The hardness testing point was specified at the center of the combustion chamber side between the intake and exhaust valve holes, which corresponded to the lowest part of the casting. To meet the hardness requirement, internal chills were placed accordingly.
During the initial validation phase, the following defects were observed:
| Defect Type | Location | Observed Frequency | Possible Direct Cause |
|---|---|---|---|
| Gas holes (choking holes) | Valve guide bores | Frequent | Moisture/gas from chills & cores |
| Pinholes | Oil gallery top surface | Moderate | Gas entrapment & poor venting |
| Shrinkage porosity | Top plane, under risers | High | Inadequate feeding, hot spots |
| Leakage | Injector holes, guide holes | 60% of total rejects | Internal shrinkage porosity |
The defects are illustrated in the original paper as: internal shrinkage in the casting, gas holes in the oil sump, choking holes in the valve guide bores, and shrinkage porosity on the top plane. The severity of these defects made the initial yield as low as 30%.
Root Cause Analysis of Gas-Related Defects
Formation of gas holes in ductile iron castings is often attributed to high moisture content in the mold or core, inadequate venting, and gas evolution from metallic chills. In our original process, the sand molds and cores were coated with alcohol-based refractory coatings. At pouring temperature, these coatings volatilize rapidly, generating a large volume of gas. Furthermore, the 45# steel internal chills, being prone to rust, introduced iron oxide and moisture on their surfaces. The coating applied on the chill surface only exacerbated the problem because the coating layer could crack and release gases. Under the high temperature of molten iron, these oxide layers decompose, releasing hydrogen and other gases that are subsequently trapped in the solidifying metal. The valve guide region, being a thick section, was particularly vulnerable because the gas bubbles could not escape easily before the metal solidified.
The oil gallery core was located at the topmost area, far from the ingates. By the time the molten iron reached this region, its temperature had dropped significantly, increasing the viscosity and reducing the ability of gas bubbles to rise and escape. The surrounding venting channels were insufficient, resulting in back-pressure and gas entrapment.
To address these gas-related defects, we implemented the following corrective measures:
1. Replacement of Steel Chills with Copper-Coated Chills
The original 45# steel internal chills in the valve guide cores were replaced with copper-plated chills. Copper has a higher thermal conductivity than steel and provides a more effective chilling action. The copper plating prevents rust formation and eliminates the source of moisture/oxide gas. Figure 4 in the original paper showed the comparison between the old and new chills. This change was critical in reducing gas evolution at the metal-chill interface.
2. Enhanced Drying of Coated Molds and Cores
The sand molds and cores, after applying the alcohol-based coating twice, were placed in a drying oven. The drying temperature was raised to 180–200 °C, and the holding time was 30 minutes. This ensured that all residual alcohol and absorbed moisture were completely removed. It was found that a single drying cycle was insufficient; the double coating with subsequent oven drying significantly reduced the gas generation during pouring.
3. Improved Venting System
Several modifications were made to the venting system:
- In the oil gallery core, a new design was introduced that provided a direct venting path to the atmosphere, as shown in Figure 5 of the original article. The modified core geometry allowed generated gases to escape more freely.
- The upper sand mold was drilled with additional vent holes (Figure 6). These vents connected the cavity to the outside, relieving pressure and allowing gas to escape.
- The air channel cores were manually drilled along their length to create internal vent channels (Figure 7). This was necessary because the cold-box cores, being dense, had no natural permeability. The drilled channels provided a path for gas to escape through the core prints.
4. Time Control After Molding
After the mold was closed (i.e., after the cores were assembled and the mold was ready), the pouring was scheduled within one hour. This prevented re-absorption of moisture from the atmosphere. In humid conditions, an even shorter interval was maintained. This practice, combined with the oven drying, minimized the risk of moisture-related gas defects.
By implementing these measures, the gas hole defects in the valve guide bores and the oil gallery pinholes were essentially eliminated. In the subsequent production lot of 200 castings, no castings were rejected due to gas holes, slag inclusions, or core drop.
Shrinkage Porosity and Leakage: Analysis and Simulation
Shrinkage porosity is a common but complex defect in ductile iron castings. The solidification of ductile iron is characterized by a wide mushy zone and a sponge-like solidification front. The formation of austenite dendrites creates a network that restricts liquid flow, making it difficult for the riser to feed the solidifying sections. Moreover, the graphite expansion during eutectic solidification can cause mold wall movement and internal stresses, leading to micro-porosity if the mold rigidity is insufficient. In the original process, despite the use of chills and insulating sleeves, shrinkage porosity was observed in the regions under the 2# and 3# risers. Hydrostatic testing of the machined cylinder heads revealed leaks at the injector holes and valve guide holes, which were traced back to internal shrinkage porosity that interconnected fine pores.
To understand the solidification behavior and to identify the locations of hot spots, we used the MAGMA simulation software. The simulation was performed on the original process to obtain the temperature distribution, thermal gradients, and hot spot evolution. The key outputs were the hot spot distribution plot and the thermal modulus curves for the risers and casting sections.
Theoretical Background: Thermal Modulus and Feeding
The thermal modulus \(M\) of a casting section is defined as the ratio of the volume \(V\) to the cooling surface area \(A\):
$$
M = \frac{V}{A}
$$
For a riser to be effective, its modulus \(M_r\) must be larger than the modulus of the casting section \(M_c\) it is intended to feed. In addition, the riser neck must not solidify earlier than the casting. The feeding distance is also influenced by the temperature gradient. In ductile iron, the graphite expansion during eutectic solidification can significantly reduce the need for external liquid feed if the mold is rigid. Chills increase the cooling rate, promote a steeper temperature gradient, and reduce the effective feeding distance required.
In the original simulation, the hot spot distribution (Figure 8 of the original article) showed:
- Under the 1# riser: the hot spot was confined inside the riser and its neck. The thermal modulus of the riser and neck exceeded that of the casting section, indicating good feeding. No shrinkage was expected in the casting.
- Under the 2# riser: the hot spot was not isolated in the riser; instead, it extended into the casting. The thermal modulus curve (Figure 9) showed that although the riser and neck moduli were higher than the casting section, a local peak in the casting modulus occurred at some distance from the riser. This peak indicated a region with a lower cooling rate, which could form an isolated hot spot that would be difficult to feed. This explains the shrinkage found in the A region.
- Under the 3# riser: the thermal modulus of the riser and neck was less than that of the casting section. Consequently, the riser solidified earlier than the casting, creating an isolated hot spot in the casting (B region). The 3# riser was located above a high, narrow thick section, and the riser neck size could not be increased due to geometric constraints.
The shrinkage porosity distribution obtained from the simulation (Figure 10) showed two main areas: A (under 2# riser) and B (under 3# riser), matching exactly the locations of leaks in the actual castings.
Process Modification with Additional Chills
Based on the simulation results, we decided to add chills in two locations of the mold, without changing the geometry of the risers. Specifically:
- On the external skin core, near the 2# riser region: two chills were added, designated as 1-1# and 1-2#.
- On the oil gallery core, near the same region: two chills were added, designated as 2-1# and 2-2#.
The philosophy behind these chills was to accelerate the cooling of the thick sections, thereby reducing the hot spot size and increasing the temperature gradient towards the riser. Chills also promote earlier graphite expansion, which helps in self-feeding of the ductile iron. The new chill arrangement is shown in Figure 11 of the original paper.
The simulation was repeated with the modified process. The results were as follows:
Hot Spot Distribution (Figure 12)
Under the 2# riser, the hot spot was now confined to the riser and neck area, with no isolated hot spot in the casting. The thermal modulus curves (Figure 13) demonstrated that the modulus of the 2# riser and neck was larger than that of the casting section, and the casting modulus continuously decreased away from the riser. This indicates a correctly directed solidification from the casting towards the riser, providing excellent feeding conditions.
3# Riser Behavior
Under the 3# riser, the hot spot size and the thermal modulus values were slightly reduced, but the improvement was not significant. The reason is that the casting structure under the 3# riser is a tall, narrow, heavy section, and the riser neck was already at its maximum allowable size. Nevertheless, the shrinkage tendency was reduced, as shown in the simulation cross-section (Figure 14). The A region shrinkage was completely resolved, while the B region showed a lower severity than before.
We also used numerical criteria to quantify the shrinkage risk. For ductile iron, the Niyama criterion is often used:
$$
N = \frac{G}{\sqrt{\dot{T}}}
$$
where \(G\) is the temperature gradient and \(\dot{T}\) is the cooling rate. A low Niyama value indicates a high risk of micro-porosity. In our simulation, the Niyama values in the A region increased significantly after adding the chills, confirming the improvement. The B region still had relatively low Niyama values, prompting further investigation for future optimization.
Production Verification and Results
After implementing the modified process, we produced a batch of 200 cylinder head castings. The results are summarized in the following table:
| Parameter | Original Process | Modified Process |
|---|---|---|
| Number of castings produced | ~100 (trial) | 200 |
| Rejections due to gas holes (valve guide) | High (frequent) | 0 |
| Rejections due to pinholes in oil gallery | Moderate | 0 |
| Rejections due to shrinkage/leakage | ~70% overall | 17 (8.5%) |
| Overall yield | ~30% | 91.5% |
The 17 rejected castings were due to top plane shrinkage and leakage. After sectioning, it was observed that the shrinkage under the 2# riser was completely eliminated. The remaining rejections were all attributed to the B region under the 3# riser. This shows that the implemented chills successfully resolved the major leakage problem, and the remaining defect is now localized and well-understood. Subsequent work is planned to further optimize the 3# riser region, perhaps by altering the casting design or using a different chill configuration.
Discussion
The results demonstrate that a systematic combination of defect analysis, simulation, and process adjustments can dramatically improve the quality of large ductile iron castings. The following points are noteworthy:
1. Gas Defects in Ductile Iron Castings
Moisture from organic binders and coatings is a major source of gas in production. The introduction of copper-coated chills not only eliminated rust but also improved the chilling action. Copper has a thermal conductivity of about 401 W/(m·K) compared to steel’s 50 W/(m·K), so it extracts heat much faster, creating a larger temperature gradient and promoting a finer microstructure. This is beneficial for both shrinkage and gas porosity. The oven drying at 180–200 °C for 30 minutes after double coating was essential. It is important to note that the alcohol-based coating should be thoroughly dried; otherwise, the residual alcohol can decompose into hydrogen. Our experience underscores the need for strict process control of core and mold drying, as well as the use of venting channels that are appropriate for the core geometry.
2. Simulation of Shrinkage in Ductile Iron Castings
MAGMA software proved to be a reliable tool for predicting hot spots and shrinkage tendencies. The thermal modulus analysis correctly identified the weak feeding condition under the 3# riser. The addition of chills to the 2# region was based on the simulated hot spot location. The simulation also helped us understand that a riser must be designed not only with a higher modulus than the casting but also with an adequate neck design. For the 2# riser, the original design had a slightly isolated hot spot in the casting; the chills eliminated this by increasing the local cooling rate and reducing the modulus peak. This is a classic example of how chills can extend the effective feeding distance of a riser.
3. Solidification Theory and Graphite Expansion
In ductile iron castings, the precipitation of graphite during eutectic solidification is accompanied by a volumetric expansion. If the mold is rigid and the skin is strong, this expansion can feed the liquid channels and reduce shrinkage porosity. Chills accelerate the cooling at critical sections, causing the graphite expansion to occur earlier in those regions, which helps to compensate for the liquid shrinkage that occurs in adjacent sections. The addition of chills in the 2# region likely shifted the temperature field such that the expansion from the cooling metal could assist in feeding. This principle is quantitatively described by the local solidification time \(\theta\):
$$
\theta = \frac{M^2}{k_\text{th}}
$$
where \(k_\text{th}\) is a constant related to the mold material. By inserting chills, we effectively reduce the local modulus \(M\) and thus shorten the solidification time, which in turn shifts the feeding behavior. However, care must be taken to avoid too rapid cooling that might lead to chill (carbide) formation. Our copper chills were coated and sized to avoid this risk.
4. Venting and Core Design
The manual drilling of vent channels in the cold-box cores was a simple yet effective solution. Since cold-box cores have very low permeability, gases generated from the binder decomposition cannot escape through the core wall. By providing central vent holes that connect to the core print, the gas path is shortened, allowing the gas to escape into the atmosphere or through the mold vents. The modification of the oil gallery core to increase its venting potential was also successful. This highlights that core geometry should not only satisfy the cavity shape but also account for gas evacuation.
5. Practical Recommendations for Large Ductile Iron Castings
Based on our experience with the L32/40 cylinder head, we propose the following guidelines for foundries producing large thick-walled ductile iron castings:
- Always use rust-free chills, preferably copper-plated, for internal chilling in critical areas such as valve guide or injector regions.
- Dry coated molds and cores thoroughly; consider a second drying cycle if the coating thickness is high.
- Design cores with built-in venting or drill vent channels after core making.
- Keep the time from molding to pouring short to avoid moisture re-absorption.
- Use simulation to identify hot spots and verify that risers have sufficient modulus. If a riser cannot be enlarged due to geometric constraints, use additional chills between the riser and the hot spot instead.
- Monitor the Niyama criterion or similar shrinkage index to evaluate the severity of micro-porosity.
Quantitative Comparison: Thermal Modulus Before and After
To further illustrate the effect of the chills, we present a simplified comparison of the thermal modulus values for the 2# and 3# riser regions. The values from Figure 9 and Figure 13 of the original paper have been normalized for the purpose of clarity. The original data indicated that the modulus of the casting section under the 2# riser had a local peak of about 3.2 cm, while after adding chills, the modulus decreased to 2.4 cm, and the riser neck modulus remained at approximately 3.8 cm. The following table gives an approximate quantitative representation:
| Region | Original Modulus (cm) | Modified Modulus (cm) | Riser/Neck Modulus (cm) | Feeding Condition |
|---|---|---|---|---|
| 1# riser | 2.8 | 2.8 | 4.0 | Adequate (unchanged) |
| 2# riser casting | 3.2 (peak) | 2.4 (no peak) | 4.2 | Poor → Good |
| 3# riser casting | 3.5 | 3.3 | 2.9 | Poor (slightly improved) |
The above table is illustrative; actual values are proprietary. The key point is that the chills reduced the casting modulus under the 2# riser, making it smaller than the riser modulus, thus achieving directional solidification.
Cheil-Zone Analysis: The Role of Chills in Solidification Control
Chills are local heat sinks that accelerate the cooling of specific regions. The effectiveness of a chill is governed by its heat capacity and thermal conductivity. For a plate-shaped chill of thickness \(t_c\), the heat absorbed can be approximated by:
$$
Q_c = \rho_c c_c V_c (T_c – T_0)
$$
where \(\rho_c\) is the density of the chill material, \(c_c\) is its specific heat, \(V_c\) is the volume of the chill, \(T_c\) is the final temperature after heat exchange, and \(T_0\) is the initial temperature. Copper, with its higher density and thermal conductivity, absorbs heat more rapidly than steel. In our application, the copper-coated chills in the valve guide cores provided a stronger chilling effect, which also helped to reduce the solidification time in those thick sections, thereby minimizing the time for gas bubbles to be trapped.
The placement of chills relative to the riser determines the temperature gradient. A desired gradient is one where the temperature decreases along the path from the riser to the remote sections. The addition of chills at the 2# region created such a gradient. We can represent the thermal gradient \(G\) as:
$$
G = \frac{T_{\text{riser}} – T_{\text{chill}}}{d}
$$
where \(T_{\text{riser}}\) is the liquidus temperature near the riser, \(T_{\text{chill}}\) is the temperature at the chill surface, and \(d\) is the distance between them. A large \(G\) promotes directional solidification and reduces the tendency for isolated hot spots. In our case, the chills increased \(G\), allowing the riser to feed more effectively.
Remaining Challenges and Future Work
Despite the significant improvement, the 3# riser region still shows a tendency for shrinkage. The root cause is the geometric constraint: the riser neck cannot be made larger because it would infringe on the casting surface. In the future, we plan to investigate alternative feeding methods, such as:
- Using an internal chill inside the 3# region, placed on the core that forms the deep cavity.
- Applying a more powerful exothermic material around the 3# riser to increase its effective modulus.
- Modifying the casting design to reduce the section thickness or to add a small pad that can be machined later.
- Exploring the use of a pressurized riser (blind riser with a gas-evolving insert) to maintain feed pressure for a longer time.
Additionally, we will continue to record pressure test results and correlate them with radiographic inspection of rejected castings to build a database for predictive quality control. The use of simulation will be expanded to evaluate the influence of pouring temperature variations and mold rigidity on shrinkage formation.
Conclusion
The production of large ductile iron castings such as the L32/40 cylinder head poses significant challenges in terms of gas defects and shrinkage porosity. Through careful analysis of the process and the adoption of specific measures—namely, copper-coated chills, enhanced drying, improved venting, and simulation-driven placement of additional chills—we achieved a dramatic reduction in the casting rejection rate from approximately 70% to 8.5%. The modifications not only eliminated the gas-related defects but also resolved the severe shrinkage under the 2# riser. The remaining defect under the 3# riser is understood and is the subject of ongoing optimization.
This case study confirms that MAGMA simulation is an invaluable tool for diagnosing and correcting casting defects in ductile iron castings. It also emphasizes that the combined use of chills and venting, guided by fundamental solidification theory, can greatly enhance the quality of complex iron castings. The lessons learned here are applicable to other large ductile iron castings where thick sections and intricate core packages are involved.
The process improvements have enabled our company to produce the cylinder heads in-house reliably, contributing to better product quality and cost savings. We continue to apply the same systematic approach to other new casting developments, ensuring that ductile iron castings meet the highest standards of integrity and performance.
This article is based on the original publication in China Foundry Machinery & Technology, 2016, No. 4.