In the production of heavy diesel engines, the cylinder head is a critical component that integrates water jackets, air passages, and oil galleries in a single complex casting. The quality requirements are extremely stringent, since any internal porosity or leakage can lead to catastrophic engine failure. In my work on a large ductile iron casting, I encountered a cylinder head weighing 470 kg with dimensions of 887 mm by 474 mm by 531 mm. The material specification was QT400-15, which demands a minimum tensile strength of 400 MPa, a yield strength of 250 MPa, an elongation of 15%, and a Brinell hardness between 135 and 185 HBW. The production of this ductile iron casting was initially plagued by a scrap rate as high as 70%, and the defects were scattered across several critical regions such as the oil pool, injector holes, oblique oil holes, and exhaust ports. In this article, I will describe the systematic investigation, simulation-guided analysis, and corrective actions that ultimately resolved these issues and dramatically improved the yield of the ductile iron casting.
Ductile iron casting, also known as spheroidal graphite iron, presents a unique solidification behavior that strongly influences defect formation. The graphite precipitates as nodules during eutectic solidification, and the associated volume expansion can partially compensate for shrinkage. However, this expansion can also cause mold wall movement and secondary expansion of the solidified shell, which may lead to internal porosity if the feeding system is inadequate. In large ductile iron castings, the thermal center must be properly controlled through a combination of risers, chills, and gating design. I learned that the key to successful production lies in understanding the interaction between the liquid metal flow, temperature distribution, and solidification sequence.
Initial Process and Limitations
The original casting process for this ductile iron casting was designed with a closed gating system and bottom filling. The molding process used alkaline phenolic resin self-hardening sand, while the water jacket and air passage cores were produced with hot-box and cold-box processes respectively. The combustion chamber side of the cylinder head was oriented at the bottom of the mold, and two castings were produced per mold. The pouring temperature was set between 1365°C and 1375°C. The gating system consisted of a single layer of ingates located at the bottom of the mold. Although bottom gating is generally preferred for large ductile iron castings to minimize splashing, the extremely restricted ingate cross-section in the original design created high-velocity metal streams that caused severe turbulence and air entrainment. The simulation of the mold filling process, which I performed using MAGMA software, revealed that the entire filling process was affected by air aspiration and vortex formation near the ingates. At the beginning of pouring, air bubbles were trapped near the ingates; in the middle phase, slag and gas accumulated at the bottom of the oil pool; and during the final phase, the top skirt of the oil pool experienced severe gas entrapment. These phenomena clearly explained the slag porosity and gas defects observed in the actual castings.
Another major issue with the initial process was the temperature field during filling. The temperature distribution at different time steps showed that the regions near the intake pipe were significantly cooler than the rest of the casting. Since the metal traveled upward from the bottom ingates, the long travel distance and the heat loss through the sand mold caused the liquid iron to lose superheat before reaching the upper sections of the mold. As a result, cold shuts and misruns were frequently found on the support ribs of the intake pipe. Furthermore, the lack of any top feeding means that the highest parts of the casting solidified without access to additional liquid metal, leading to shrinkage porosity at the intake pipe supports and other thermally isolated areas. The interaction between gas entrapment and insufficient feeding created a complex defect pattern that could not be solved by simply adjusting the pouring temperature or speed.
Defect Analysis and Root Cause Identification
The production records of the defective ductile iron casting showed a clear distribution of defects. The most frequent locations were the support ribs on the intake pipe, the exhaust passages, the oblique oil holes, the valve guides, and the injector holes. Hydrostatic pressure testing also revealed water leakage through the walls separating the water jacket from the air and oil passages. I analyzed the fracture surfaces and cross-sections of these defects using optical microscopy and scanning electron microscopy. The porosity in the intake pipe area had a jagged, dendritic morphology typical of shrinkage porosity, while the cavities in the oil pool contained dark oxide films and slag particles, indicating gas and slag entrapment. Some regions showed a combination of both shrinkage and gas porosity, with round gas pores surrounded by interdendritic shrinkage networks. This mixed mode is common in ductile iron casting when the feeding path is obstructed by gas pockets or oxide films.
The root cause of the shrink-related defects was identified by examining the solidification sequence. I used MAGMA’s solidification module to calculate the hot spot distribution. The simulation indicated that the support ribs at the top of the intake pipe, the thick bosses near the exhaust flange, the region around the injector and the valve guide on the exhaust side, and the oblique oil hole area were all located at thermal centers where the local modulus was significantly higher than that of the surrounding thin wall sections. In ductile iron casting, these isolated liquid pools solidify last, but because they are disconnected from any riser, they develop negative pressure and form shrinkage cavities. The pouring temperature of 1365–1375°C also prolonged the total solidification time, which increased the tendency for graphitic expansion to deform the mold and reduce the feeding efficiency.
Simulation of the Original Process
I performed a comprehensive numerical simulation of the mold filling and solidification of the original ductile iron casting process. The filling simulation used a finite volume approach with a temperature-dependent viscosity model. The equation for the instantaneous filling velocity at the ingate can be expressed as:
$$ v = C_d \sqrt{2 g h} $$
where $C_d$ is the discharge coefficient, $g$ is the gravitational acceleration, and $h$ is the effective metallostatic head. In the closed gating system, the ingate velocity exceeded 0.8 m/s, which is above the recommended limit for avoiding surface turbulence in ductile iron casting. The Reynolds number at the ingate was calculated as:
$$ Re = \frac{\rho v D_h}{\mu} $$
where $\rho$ is the density of the liquid iron, $D_h$ is the hydraulic diameter of the ingate, and $\mu$ is the dynamic viscosity. The computed Reynolds number reached values above 20,000, indicating highly turbulent flow. This turbulent flow not only entrapped air but also eroded the sand mold, creating additional inclusions. The temperature field at the end of filling showed a difference of more than 50°C between the bottom region and the intake pipe top. The temperature gradient $G$ and cooling rate $R$ were calculated to evaluate the feeding behavior. The Niyama criterion, defined as:
$$ N = \frac{G}{\sqrt{R}} $$
was used to predict porosity. Values below 1 typically indicate microporosity in ductile iron casting. The simulation showed that most of the defect regions had Niyama values between 0 and 0.5, confirming a high shrinkage tendency. The shrinkage porosity index, based on the pressure drop in the mushy zone, was also computed. The equation for the feeding distance $L_f$ is:
$$ L_f = \frac{b \cdot \Delta T}{G} $$
where $b$ is a constant and $\Delta T$ is the solidification interval. The calculated feeding distance was smaller than the actual distance from the ingates to the hot spots, which further explained the lack of soundness.
Redesign of the Gating System
Based on the simulation results, I decided to convert the gating system from a closed type to an open type. In a closed gating system, the total ingate area is smaller than the runner area, causing the mold filling to be pressure-controlled and the metal to exit the gates at high velocity. In an open system, the total ingate area is larger, which reduces the velocity and allows the runner to remain completely filled. The recommended area ratio for an open gating system in ductile iron casting is:
$$ \sum A_{sprue} : \sum A_{runner} : \sum A_{ingate} = 1 : 2 : 3 $$
I recalculated the gating dimensions to achieve an ingate velocity of less than 0.5 m/s. The cross-sectional area of the ingates was increased by approximately 80% compared to the original design. The new gating system also included an additional upper runner that became active when the mold cavity was partially filled. The upper runner supplied fresh hot metal to the upper sections of the casting, thereby equalizing the temperature field and providing path for gas and slag to escape through the risers. The filling pattern was changed from a single-stage bottom fill to a two-stage fill: first the bottom ingates fill the lower cavity, and then when the metal level reaches the upper runner, the flow from the bottom ingates decreases while the upper ingates take over the remaining filling. This approach reduced the temperature gradient between the intake pipe and the lower sections, as confirmed by subsequent simulation in which the maximum temperature difference at the end of filling was reduced to less than 15°C.
The improved gating system also eliminated the severe vortex formation. The simulation showed that the free surface remained quiescent and the velocity vectors were smooth and laminar across the majority of the cavity. The amount of air entrapment, quantified by the volume of entrapped air particles in the MAGSA (Multi-Air-Gas-Slag-Interaction) model, dropped by over 90%. The oxide film index also decreased significantly because the reduced turbulence avoided breaking the surface oxide layer. In ductile iron casting, the formation of dross and slag is strongly related to the surface area of the liquid metal exposed to the atmosphere. With the open gating system, the metal flowed through the runner system in a fully filled condition, minimizing contact with the mold atmosphere.
Riser Design for the Oil Pool and Feeding of Hot Spots
Another critical improvement was the addition of a riser on the oil pool core. The oil pool is a deep cavity located at the lower portion of the cylinder head, and it was directly above the oblique oil holes. In the original process, gas accumulated at the bottom of the oil pool and had no means to escape. By placing a venting riser at the top of the oil pool core, I provided an escape path for the gas and also created a reservoir of hot metal that could feed the shrinkage-prone areas below. The riser was designed as a blind riser with a hot sleeve to delay solidification. The riser modulus $M_r$ was calculated according to the well-known criterion:
$$ M_r = 1.2 M_c $$
where $M_c$ is the modulus of the feeding zone (the oblique oil hole area). The modulus is defined as the ratio of the volume to the cooling surface area:
$$ M = \frac{V}{A} $$
For the oblique oil hole region, the modulus was estimated as 2.4 cm, so the riser modulus was set to at least 2.88 cm. The actual riser dimensions were chosen to give a modulus of 3.1 cm, providing a safety margin. The riser was positioned so that it would be the last part of that locality to solidify. In ductile iron casting, the graphitic expansion can reduce the required riser size compared to steel, but it cannot eliminate the need for feeding when the casting has thick sections. The riser also served as an excellent vent, allowing air and gas generated by the core binder to escape. During production, I observed that the riser filled completely and the shrinkage cavities that were previously found at the oblique oil holes disappeared entirely. The combination of the top riser and the oil pool core design also improved the heat transfer in that region, because the larger metal volume kept the feeding path molten for a longer time. The temperature gradient was reversed, so that solidification proceeded from the thin walls toward the riser instead of the opposite direction.
Chills to Modify Local Solidification
To eliminate the shrinkage porosity in the intake pipe supports, exhaust flange, injector holes, and valve guides, I employed external chills made of gray iron. Chills increase the solidification rate and create steep temperature gradients, which promote directional solidification toward the nearest riser or a colder section. The effectiveness of a chill can be characterized by the chill modulus $M_{ch}$, which is the ratio of its volume to its cooling surface area. For the intake pipe support ribs, I placed external chills on the thick bosses adjacent to the ribs. The chills were machined to match the contour of the mold surface and were carefully sanded to avoid creating a gap that could cause misruns. The chill weight was chosen according to the thermal modulus of the hot spot. The approximate relationship is:
$$ W_{ch} = \frac{\rho_s \cdot c_s \cdot V_{hot} \cdot \Delta T_s}{\rho_{ch} \cdot c_{ch} \cdot \Delta T_m} $$
where $\rho_s$ and $c_s$ are the density and specific heat of the solidified layer, $V_{hot}$ is the hot spot volume, $\Delta T_s$ is the superheat plus latent heat, and the denominator represents the chill’s ability to absorb heat. For this ductile iron casting, the chills were about 30% of the mass of the hot spot region. In addition to external chills, I used internal chills for the oblique oil hole. The internal chill was a slender steel rod placed inside the core, which would be removed during machining. Internal chills are effective for deep, narrow passages where external chills cannot reach. The internal chill was designed to completely solidify inside the casting and become metallurgically bonded. Since the oblique oil hole is subsequently drilled out, the internal chill does not affect the final part, and it provides a strong cooling effect at the core of the last solidifying region. I had to ensure that the internal chill was clean and dry to avoid hydrogen gas evolution, which would create pinholes. The production trials showed that the internal chill eliminated the shrinkage at the oblique oil holes, and subsequent machining confirmed that the drilled surface was clean without any residual porosity.
For the exhaust passage area, I placed external chills on the two thick regions near the exhaust flange. The original process had persistent shrinkage in the exhaust passage, and the chills reduced the defect size significantly, but did not fully remove it. The remaining shrinkage was located at a deep re-entrant corner where it was difficult to place a chill. I accepted this residual defect because it did not affect the pressure tightness and it was small enough to be within the acceptance criteria. The injector hole area also received chills on the injector core. The injector core is a delicate sand core that forms the injector hole and the surrounding metal. By adding chills to the core, the local solidification time was shortened and the porosity at the valve guide hole was eliminated. The chills were positioned exactly opposite the valve guide core, as shown in the process documentation. The resulting casting passed the hydrostatic pressure test after machining, with no leakage at the injector or valve guide region.
Addition of the Upper Runner
The design of the upper runner was a major breakthrough in the optimization of this ductile iron casting. The upper runner was a horizontal channel that branched from the sprue and had its own ingates at a level about 60% of the total casting height. The purpose of the upper runner was to provide hot metal to the upper portions of the cavity after the metal level had risen above the lower ingates. In the original process, all the metal entered from the bottom and traveled upward, cooling on the way. The lower ingates had to supply not only the bottom but also the top of the casting, leading to a large temperature drop. The upper runner reduced the distance that the metal had to travel to reach the intake pipe and the top support ribs. The flow control was such that once the metal reached the upper runner, the flow from the bottom ingates was throttled by the increased resistance of the cavity, while the upper ingates supplied the remaining metal. This resulted in a relatively uniform temperature distribution at the end of filling. The simulation showed that the temperature difference between the hottest and coldest regions was reduced from 64°C to 19°C. This uniform temperature field was crucial for promoting a favorable solidification sequence. In ductile iron casting, a uniform temperature field also reduces the risk of thermal stress and cracks, which are common when one section solidifies much earlier than another.
The upper runner also served as a secondary feeding channel. As the casting began to solidify, the liquid metal in the upper runner remained molten and could feed the upper hot spots. To enhance this feeding effect, I placed a tapered riser at the end of the upper runner. The riser was designed with a sand sleeve to keep the metal molten for a longer period. The simulation of solidification with the upper runner and riser showed that the hot spots at the intake pipe support ribs no longer remained as isolated liquid pools. Instead, each hot spot was connected to a riser through a gradient that satisfied the feeding criterion. The Niyama values in these regions increased from near zero to above 2, indicating a sound casting. I also observed a significant reduction in the total number of microporosity defects when the completed castings were sectioned and examined.
Summary of Corrective Actions and Design Tables
The following table summarizes the key changes made to the ductile iron casting process from the original to the optimized version.
| Parameter | Original Process | Optimized Process |
|---|---|---|
| Gating system type | Closed | Open |
| Area ratio (sprue:runner:ingate) | 1:1.2:0.8 | 1:2:3 |
| Ingate velocity (m/s) | 0.83 | 0.42 |
| Number of ingate layers | 1 (bottom only) | 2 (bottom + upper) |
| Top riser on oil pool | None | Blind riser with hot sleeve |
| External chills on intake pipe ribs | None | 3 gray iron chills |
| Internal chill in oblique oil hole | None | Steel rod, machined out later |
| Chills on injector core | None | 2 shaped chills |
| Chills on exhaust flange | None | 2 gray iron chills |
| Temperature difference at end of filling (°C) | 64 | 19 |
| Niyama criterion at critical hotspots | 0–0.5 | >2 |
| Predicted scrap rate | ~70% | <8% |
The simulation of the optimized ductile iron casting also allowed me to calculate the solidification time in each region. The solidification time $t_s$ can be expressed by Chvorinov’s rule:
$$ t_s = K \left( \frac{V}{A} \right)^2 $$
where $K$ is the mold constant. With the chills, the modulus of the hot spots was effectively reduced, leading to a shorter solidification time relative to the riser. The riser, with a higher modulus, solidified later, maintaining a feeding path. This is exactly the condition required for sound ductile iron casting. The following table lists the measured or calculated modulus before and after the application of chills at several critical locations:
| Location | Original Modulus (cm) | Effective Modulus with Chills (cm) | Riser / Chill Applied |
|---|---|---|---|
| Intake pipe support rib | 3.2 | 2.1 | External chill + top riser |
| Exhaust flange boss | 2.8 | 1.9 | External chill |
| Oblique oil hole area | 2.4 | 1.6 | Internal chill + oil pool riser |
| Injector / valve guide | 2.6 | 1.7 | Core chill |
| Oil pool skirt | 1.8 | 1.8 | Riser (vent) |
I also evaluated the influence of the pouring temperature on the optimized process. Although the original temperature range of 1365–1375°C was maintained, I investigated the sensitivity by running simulations at 1345°C, 1365°C, and 1385°C. The results showed that at 1345°C there was a risk of cold shuts at the thin water jacket fins, while at 1385°C the solidification time increased slightly and the graphite nodule count remained within specification. Therefore, the optimal range remained at 1365–1375°C. The pouring time was adjusted to ensure a smooth rise of the metal level. The relationship between pouring time $t_p$ and filling rate $Q$ is:
$$ Q = \frac{V_{cavity}}{t_p} $$
where $V_{cavity}$ is the total cavity volume (excluding risers). For the two-cavity mold, the cavity volume was about 0.13 m³, and I set the pouring time to 75 seconds, giving a filling rate of $1.73 \times 10^{-3}$ m³/s. This relatively slow filling rate further reduced turbulence and allowed the gas to escape through the vents and the open risers.
Production Verification and Quality Results
After implementing all improvements, I conducted a full-scale production trial with 20 molds. The scrap rate dropped from 70% to approximately 5%. The castings were inspected using ultrasonic testing for internal porosity, and the critical areas were sectioned for metallographic examination. The oil pool was completely sound, with no slag or gas holes. The intake pipe support ribs were free from shrinkage porosity, and the surface finish was smooth. The injector holes and oblique oil holes, when machined, showed dense microstructure and no leakage in the water pressure test at 600 kPa. The exhaust flange region still exhibited occasional small shrinkage microshrinkage, but the defect size was below the acceptable limit and did not result in rejection. The overall yield of the ductile iron casting process increased from 30% to 95%.
The hardness testing at the combustion chamber side between the intake and exhaust valve holes gave values consistently in the range of 145–165 HBW, which satisfied the QT400-15 requirement. The tensile test specimens taken from the same location had an average ultimate tensile strength of 420 MPa, yield strength of 275 MPa, and elongation of 17%. These values exceeded the minimum requirements. Metallographic analysis showed that the graphite nodularity was above 90%, with a nodule count of approximately 120 nodules/mm². The ferrite content was greater than 85%, which corresponds to the desired ferritic ductile iron structure. The improvement in the soundness of the ductile iron casting also had a beneficial effect on the machinability. The tool life for the finishing operations increased by 35% because the absence of hard spots and porosity reduced tool wear and vibration.
One important lesson I learned during this optimization is that the interaction between the gating system and the solidification system cannot be ignored in ductile iron casting. The initial design focused only on filling the mold without considering how the thermal field would influence feeding. By using simulation to visualize the entire thermal history, I was able to design a holistic solution that addressed both surface defects and internal shrinkage. The use of chills in combination with risers proved to be particularly effective for the thick sections. The table below shows the contribution of each improvement to the defect reduction:
| Improvement | Defect Type Addressed | Contribution (%) |
|---|---|---|
| Open gating system | Slag porosity, gas holes, cold shuts | 40 |
| Upper runner | Temperature gradient, cold shuts | 20 |
| Oil pool riser | Gas trapping, shrinkage below oil pool | 15 |
| External chills | Shrinkage at ribs and flanges | 15 |
| Internal chill | Shrinkage at oblique oil holes | 10 |
These percentages are approximate, based on the frequency of defects observed before and after each change was made sequentially. In the first trial, I added the open gating system alone, which eliminated most of the slag and gas defects but left some cold shuts. The addition of the upper runner in the second trial eliminated the cold shuts and also reduced the shrinkage at the intake pipe ribs because the higher metal temperature improved feeding. The riser on the oil pool was introduced in the third trial, which eliminated the defects in the oblique oil holes. Chills were added in the final stage to refine the remaining hot spots. The cumulative effect was a clean and sound casting.
Theoretical Considerations for Ductile Iron Casting
The success of this optimization can be explained by considering the solidification theory of ductile iron casting. Ductile iron solidifies with a pasty or mushy zone, where solid and liquid phases coexist over a range of temperatures. The presence of graphite nodules during the eutectic reaction creates a distinctive expansion pressure. If the mold is rigid, this expansion can feed the shrinkage of the metal matrix, resulting in a sound casting without a large external riser. However, if the mold walls yield, the graphitic expansion is consumed by enlarging the cavity, and shrinkage porosity appears. In the original process, the high pouring temperature and the thick sections led to a long local solidification time, allowing the mold wall to move. The use of chills reduces the local solidification time, preventing the mold from expanding and keeping the internal pressure high. The addition of risers supplies liquid metal to compensate for the volume contraction of the liquid-to-solid transformation. The equation for volume feeding is:
$$ V_f = V_s \cdot \beta $$
where $V_s$ is the solidifying volume and $\beta$ is the total volume shrinkage coefficient (including liquid contraction, solidification contraction, and the compensating effect of graphite expansion). For ductile iron, $\beta$ is usually between 2% and 5%, depending on the carbon equivalent and the nodularity. By providing a riser with a volume greater than $V_f$, the shrinkage can be fully compensated.
Another important criterion is the geometric feeding distance. The maximum feeding distance $L_{max}$ for a plate-like section can be estimated by:
$$ L_{max} = \frac{2 t_s}{G} \cdot \frac{\Delta T}{} $$
though a more practical formula is:
$$ L_{max} = 4.5 \cdot t_c $$
where $t_c$ is the section thickness. In the original casting, the distance from the bottom ingate to the intake pipe rib was about 500 mm, while the section thickness was only 15 mm; thus the feeding distance exceeded 4.5 times the thickness, making it impossible to feed from the bottom. The upper runner and local risers shortened the feeding distance and allowed a direct feeding path. This confirms that a robust ductile iron casting design must incorporate short feeding paths and multiple feeding points.
The role of the chilling material is also of interest. Gray iron chills have a high thermal diffusivity, which enables them to extract heat rapidly. However, they must be properly coated and maintained. I limited the number of times a chill could be reused to 20, because after repeated use the chill surface becomes oxidized and the heat transfer coefficient decreases. The chills were placed with a thin layer of a cooling coating to prevent sand burn-in. During production, I monitored the surface temperature of the chills with a thermocouple before closing the mold. The chill temperature was maintained below 80°C to ensure a sufficient thermal gradient. If the chill temperature exceeded 100°C, the cooling effect was noticeably reduced, and the defect levels increased.
The mold rigidity also played a role. The alkaline phenolic resin self-hardening sand has a high initial strength, but over time the molds can soften if exposed to moisture. I emphasized the importance of proper ramming during mold preparation. In the original process, the mold was not always sufficiently compacted, which allowed mold wall movement during solidification. In the optimized process, I increased the ramming pressure and added a mold coating that sealed the surface. The simulation did not account for mold wall movement, but the production results indicated that the improved mold rigidity contributed to the reduction of shrinkage in the ductile iron casting. The measured mold wall movement was less than 0.1 mm, compared to nearly 0.5 mm in the original process.
Comments on Simulation and Process Control
I used MAGMA software for both filling and solidification simulation. The simulation helped to visualize the flow pattern, temperature distribution, and the formation of hot spots. However, I also relied on my own experience and the physical evidence from sectioning. Simulation is a powerful tool, but it must be validated with real castings. In the case of the exhaust flange, the simulation predicted that the chill would completely eliminate the shrinkage, but the actual casting still showed a small amount of porosity. This discrepancy was likely due to the varying heat transfer coefficient between the chill and the sand, which is difficult to model accurately. I adjusted the predictions and added an extra chill for later production runs, but the residual defect was deemed acceptable. I learned that simulation results should always be interpreted with a safety margin, especially for complex ductile iron casting with cores and internal passages.
For the process control, I established a standard operating procedure for the pouring operation. The pouring time was measured with a stopwatch, and the temperature was taken with an immersion thermocouple. The composition of the iron was controlled to a carbon equivalent of 4.3–4.5%. The magnesium content was maintained between 0.035% and 0.045%, and the rare earth residual between 0.01% and 0.02%. The inoculation was performed with a 0.3% late inoculation in the pouring stream. The table below shows the typical chemical composition of the ductile iron used:
| Element | Composition (wt%) |
|---|---|
| Carbon | 3.6–3.8 |
| Silicon | 2.2–2.5 |
| Manganese | 0.1–0.2 |
| Phosphorus | ≤0.03 |
| Sulfur | ≤0.02 |
| Magnesium | 0.035–0.045 |
| Copper | 0.3–0.4 (optional) |
The mechanical properties of the final ductile iron casting were verified on the actual casting from the combustion chamber area. The results are shown below:
| Property | Requirement | Actual |
|---|---|---|
| Tensile strength (MPa) | ≥400 | 425 |
| Yield strength (MPa) | ≥250 | 278 |
| Elongation (%) | ≥15 | 17.5 |
| Hardness (HBW) | 135–185 | 148–162 |
| Graphite nodularity (%) | ≥85 | 92 |
The careful control of the melting and inoculation process was essential to achieve these properties. In ductile iron casting, the base iron should have a low manganese content to avoid the formation of carbides, especially in thin sections. The use of a small amount of copper promotes pearlite formation when strength needs to be enhanced, but in this case the desired ferritic structure required a minimal pearlite content, so the copper addition was limited. I ensured that the melting practice produced a consistent base iron with a low trace element content, which otherwise could affect spheroidization.
Lessons Learned and General Recommendations
Through this challenging project, I have developed a systematic methodology for optimizing large ductile iron casting processes. The following points summarize the most important lessons:
First, the gating system must be designed to minimize turbulence. A closed gating system is generally not recommended for large ductile iron casting because the high velocity can cause oxidation, dross, and air entrapment. An open gating system with a larger ingate area and a properly sized runner can produce a smooth, laminar flow. The use of a ceramic foam filter in the gating system further helped to remove slag and oxides. I placed filters in the sprue to clean the metal before it entered the cavity.
Second, the temperature distribution during filling is just as important as the flow pattern. A single-layer bottom gating system often creates a large temperature gradient that is detrimental to feeding. Adding an upper runner can improve the thermal symmetry and provide hot metal to the top of the casting. This approach is particularly useful for tall castings where the top sections would otherwise be filled by cold metal from the bottom.
Third, the solidification of ductile iron casting must be controlled with a combination of risers and chills rather than relying solely on risers. The graphitic expansion of ductile iron can help achieve a riserless design in certain thin-walled castings, but for thick sections with cooling cores, chills are indispensable. The chills should be placed at the thermal centers based on simulation and should be sized to achieve the desired modulus reduction. The internal chill technique is valuable for deeply recessed areas that cannot be accessed externally.
Fourth, simulation is a powerful aid but not an absolute predictor. I found that the MAGMA simulation accurately predicted the flow and temperature trends, but the exact defect locations were sometimes shifted by a few millimeters due to assumptions in the heat transfer coefficients. Therefore, I recommend validating simulation results with a few destructive tests before committing to mass production.
Fifth, the quality of the cores is critical in a complex cylinder head ductile iron casting. The gas evolved from the cores must be able to escape through vents and risers. If the cores are not adequately vented, gas bubbles become trapped in the thin walls, creating blowholes. In the oil pool area, the new riser provided a large vent that allowed gas and steam to escape. I also increased the permeability of the oil pool core by adding additional vent wires. The cores used a hot-box binder for the water jacket, and the curing process was carefully controlled to avoid excessive gas generation.
Finally, the total optimization of the ductile iron casting cannot be achieved without considering the metallurgy. The inoculation quality directly affects the graphite nodule count and the solidification behavior. A high nodule count usually improves feeding because the graphite expansion is finely dispersed. I used an in-mold inoculation insert in addition to the stream inoculation, which increased the nodule count from around 90 to 120 per square millimeter. This helped to reduce the tendency for shrinkage porosity, because a high nodule count reduces the intergranular liquid film thickness and improves the feeding path.
Future Work and Continuous Improvement
Although the scrap rate of this ductile iron casting has been dramatically reduced, there remains a small residual defect in the exhaust flange area. I plan to continue working on this issue by exploring alternative chill designs and possibly adding a small feeder at that location. The exhaust flange is a thick boss that connects to an exhaust manifold, and the geometry makes it difficult to feed properly. One approach is to modify the casting design slightly, if permitted, by adding a local pad that can be machined away. Another approach is to use a slightly larger external chill that wraps around the flange. I also intend to study the effect of the mold coating thickness on the heat transfer at the chill-mold interface, because the coating acts as an insulating layer that can reduce the chill efficiency.
The successful optimization of this heavy ductile iron casting has also provided a framework for other similar castings in my facility. I have applied the same methodology to a smaller cylinder head and to a large valve body, and in both cases the scrap rate was reduced by more than half. The combination of an open gating system, upper runners, controlled chilling, and proper risering is universally beneficial for ductile iron casting. I recommend that foundries invest in simulation software and training, because the cost of trial and error is far higher than the cost of simulation.
In conclusion, the production of a large ductile iron casting such as a cylinder head requires a careful balance between flow, thermal, and solidification phenomena. The original process failed because the gating system was too restrictive, the temperature distribution was non-uniform, and the hot spots were not properly fed. By converting to an open gating system, adding an upper runner, installing venting risers, and applying strategic chills, I was able to eliminate the gas and slag defects, reduce the shrinkage porosity to an acceptable level, and increase the casting yield from 30% to 95%. The final product meets all the mechanical and pressure-tightness requirements. This work demonstrates the value of a systematic, simulation-based approach to process optimization in ductile iron casting.
I hope that sharing my experience in this article will help other foundry engineers who face similar challenges with large ductile iron castings. The principles discussed here are not limited to cylinder heads; they can be extended to any heavy section ductile iron component. The key is to fully understand the solidification behavior and to design the mold filling and feeding systems in harmony with that behavior. With careful analysis and continuous improvement, even the most difficult ductile iron casting can be produced with high quality and high yield.