In our foundry, the production of heavy ductile iron castings has always been a challenging task due to the complex solidification behavior and the stringent quality requirements for internal soundness. One of the most demanding components we manufacture is a large cylinder head for a diesel engine. This cylinder head is not only a typical example of heavy ductile iron castings but also a critical safety-related part that must withstand high thermal and mechanical loads during service. The component integrates water jackets, air passages, and oil galleries in a single casting, and any internal defect such as shrinkage porosity, gas holes, or slag inclusions can lead to leakage and eventual failure. In this article, I will share our experience in improving the casting process for this heavy ductile iron cylinder head, focusing on the root cause analysis, numerical simulation, and the implementation of effective corrective actions.
The cylinder head we produce is made of QT400-15 ductile iron, which requires a minimum tensile strength of 400 MPa, a yield strength of 250 MPa, and an elongation of 15%. The hardness in the specified area, located between the intake and exhaust valve seats on the combustion chamber side, must be between 135 and 185 HBW. The overall dimensions of the casting are 887 mm × 474 mm × 531 mm, with a finished weight of about 470 kg. The casting complexity is high because it includes an integrated intake manifold and thin-walled water jacket cores. Initially, the scrap rate during the validation phase was as high as 70%, which was unacceptable for economic production. The typical defects observed were shrinkage porosity in the intake pipe support ribs, cold shuts, shrinkage in the exhaust passages, shrinkage in the oblique oil hole, valve guide bores, and injector holes, as well as water leakage under hydrostatic testing. These defects are common in heavy ductile iron castings, especially when the casting design has thick sections adjacent to thin walls, creating isolated hot spots.
To systematically address these issues, we first reviewed the original casting process. The combustion chamber face was placed at the lowest point of the mold, and a bottom gating system with two cavities per mold was adopted. The molds were made of alkaline phenolic resin self-hardening sand, and the water jacket cores were produced using a hot box process, while the remaining cores used triethylamine cold box technology. The pouring temperature was controlled between 1365 and 1375°C. A three-dimensional representation of the casting is shown in the figure below.
The initial gating system was designed as a closed system with a relatively small ingate cross-section. During mold filling, the liquid metal experienced severe turbulence and splashing, which entrained air and oxidized dross into the melt. The simulation of the filling process using MAGMA software clearly revealed that the entire cavity was subjected to air entrapment. Early in the pouring, gas was trapped near the ingates; during the middle stage, gas accumulated at the bottom of the oil sump; and at the final stage, severe air entrapment occurred around the skirt of the oil sump. Moreover, the temperature field during filling was highly non-uniform. The regions near the intake pipe remained at a lower temperature, which increased the risk of cold shuts and misruns. The closed gating system with a single bottom ingate did not provide adequate feeding of the upper parts of the casting, leading to premature solidification of the metal in those areas and the inability to expel gas and inclusions.
The shrinkage defects, on the other hand, were attributed to the unique solidification behavior of ductile iron castings. Ductile iron solidifies with a mushy or pasty mode, where the primary austenite dendrites and graphite nodules grow simultaneously over a wide temperature range. The graphite precipitation during eutectic solidification causes an internal expansion force, which can compensate for solidification shrinkage if the mold is rigid enough. However, in a sand mold with limited rigidity, the expansion can cause mold wall movement, leading to internal porosity. The feeding behavior of ductile iron castings is further complicated by the fact that the liquid-to-solid transformation is not sharp, and isolated liquid pools can form in hot spots. Our thermal analysis of the casting identified several critical hot spots: the support ribs at the top of the intake pipe, the heavy bosses near the exhaust flange, the area around the injector and the exhaust-side valve guide bore, and the oblique oil hole. These regions were the last to solidify and were not adequately fed by the gating system or risers.
To better understand the solidification sequence, we performed a detailed thermal simulation. The temperature gradient in these hot spots was extremely low, resulting in a pasty zone that persisted for a long time. The simulation results matched the actual defect locations very well. Based on this analysis, we concluded that the shrinkage defects in heavy ductile iron castings could be mitigated by combining chills to increase the cooling rate and promote directional solidification, with risers to provide liquid metal feeding and to allow gas and inclusions to escape. The use of insulating risers is particularly effective for ductile iron castings because they maintain a liquid reservoir for a longer time, compensating for the volumetric shrinkage that occurs during the austenite liquidus-to-solidus transformation.
In the next step, we redesigned the gating system to achieve a more gentle and controlled filling of the mold cavity. The cross-section ratio of the sprue, runner, and ingates was changed from a closed to an open system. A typical open gating system has a ratio such as \(A_{sp}:A_{ru}:A_{in} = 1:2:4\) or similar, where the total cross-sectional area increases progressively. This ensures that the runner is always full but the ingates are not, reducing the velocity of the metal entering the cavity. The new ingates were made larger in cross-section and positioned to minimize the free-fall height of the molten metal. Additionally, we added an upper runner system that would begin to feed the cavity once the melt level reached that height. This upper runner helped to maintain a higher temperature in the upper regions of the casting, especially near the intake manifold, and facilitated the removal of gases and inclusions that would otherwise become trapped.
For the oil sump area, we incorporated a riser on the oil sump core. This riser served a dual purpose: it acted as a vent to release the gases that had accumulated under the core, and it provided a feeding path for the oblique oil hole region located below. A schematic of this riser arrangement is shown in Table 1, where we summarize all the improvements applied to each critical area.
| Defect Location | Root Cause | Corrective Action | Expected Effect |
|---|---|---|---|
| Intake pipe support ribs | Cold shut, shrinkage due to low temperature and hot spot | Added top riser on intake pipe, applied chills on the bosses | Improved feeding, accelerated cooling, eliminated shrinkage |
| Exhaust passage | Shrinkage at hot spot near flange | Applied chills on the hot spot area | Reduced shrinkage, improved local solidification |
| Injector hole and valve guide bores | Concentrated hot spot | Added chills on the injector core, aligned with the valve guide bores | Enhanced cooling, reduced porosity |
| Oblique oil hole | Isolated liquid pool, last to solidify | Added insulating riser on top, inserted internal chill in the oil hole area | Provided feeding, accelerated solidification |
| Overall filling turbulence | Closed gating system, splashing and air entrapment | Changed to open gating system, enlarged ingates, added upper runner | Smooth filling, reduced gas entrapment, uniform temperature |
| Oil sump gas entrapment | Gas trapped under core | Added riser on oil sump core | Improved venting and gas escape |
One of the key theoretical considerations for the successful production of heavy ductile iron castings is the design of chills and risers based on the thermal modulus of the casting sections. The local solidification time can be estimated using Chvorinov’s rule, which states that the solidification time \(t\) is proportional to the square of the volume-to-surface-area ratio (modulus \(M\)) of the casting section:
$$ t = B \left( \frac{V}{A} \right)^2 = B M^2 $$
where \(B\) is a mold constant that depends on the mold material and the thermophysical properties of the metal and mold. For a hot spot with modulus \(M_c\), a chill can reduce the local solidification time by increasing the heat extraction rate. The equivalent modulus of a chilled section can be approximated by the ratio of the volume to the effective cooling surface area, which includes the chill surface. We used this principle to determine the size and location of the chills in the areas prone to shrinkage. For instance, at the intake pipe bosses, we placed graphite chills with a thickness approximately equal to the section thickness to achieve a steep temperature gradient.
Riser design for ductile iron castings also requires careful consideration. Unlike steel castings, ductile iron castings can be fed more effectively by utilizing the graphitic expansion during eutectic solidification. However, a riser is still necessary to compensate for the liquid shrinkage before the eutectic reaction and to provide a path for gas to escape. The required riser volume \(V_r\) can be estimated from the total shrinkage of the casting, taking into account the liquid shrinkage and the solidification shrinkage. The total volumetric contraction for ductile iron is often considered to be around 3% to 4% if the mold expands, but with a rigid mold, the expansion can partially compensate. In practice, we used insulating risers with a height-to-diameter ratio of about 1.5, and the riser neck was designed to remain liquid until the casting section had solidified. The standard feeding distance for ductile iron in a plate-like section is given by:
$$ L_f = 2M + 4M = 6M $$
where \(M\) is the modulus of the plate. For more complex geometries, we relied on simulation results to optimize the riser placement.
The internal chill inserted in the oblique oil hole was a unique measure. Since the oil hole is a long slender passage that is later machined, we placed a thin steel rod of small diameter into the core so that after casting it would be completely surrounded by metal and subsequently removed during machining. This internal chill acted as a heat sink, drawing heat from the center of the hot spot and effectively eliminating the shrinkage porosity that had been observed in that region. The use of internal chills in heavy ductile iron castings must be carefully controlled to avoid defects such as lack of fusion or porosity around the chill, but with proper cleaning and coating, this approach proved to be very effective.
In addition to the localized measures, we also optimized the pouring practice. The pouring temperature was reduced slightly to 1355–1365°C to minimize the risk of mold expansion and to refine the graphite structure. A lower pouring temperature also reduces the total heat input, which helps in achieving a more favorable temperature gradient. However, we had to ensure that the temperature was still high enough to avoid cold shuts and misruns, especially in the complex thin-walled sections. The simulation showed that the temperature distribution at the end of filling became more uniform with the new gating system, with a maximum temperature difference of less than 25°C across the entire casting, compared to over 60°C in the original process.
The production verification was carried out on a batch of 20 castings. The first improvement we observed was the complete elimination of slag and gas holes. The subsequent machining and hydrostatic testing revealed that the shrinkage defects in the intake pipe area had disappeared, and the injector holes and oblique oil holes showed only minor unacceptable porosity in very few samples. The overall scrap rate dropped from 70% to less than 15%, and most of the remaining scrap was due to residual shrinkage at the exhaust flange area, which was less severe and did not affect the pressure tightness of the component. We continued to work on that area by adding an additional chill and a small venting riser, but the results from the initial improvements were already very satisfactory.
To further illustrate the effectiveness of the improved process, we compared the defect occurrence before and after the changes. Table 2 presents a summary of the defect rates for the major categories.
| Defect Type | Before Optimization (%) | After Optimization (%) |
|---|---|---|
| Slag and gas holes | 18 | 0 |
| Shrinkage in intake pipe | 25 | 0 |
| Shrinkage in exhaust passage | 15 | 8 |
| Shrinkage in injector and valve guide holes | 10 | 2 |
| Shrinkage in oblique oil hole | 8 | 1 |
| Water leakage (hydrostatic test) | 20 | 3 |
The improvement in the quality of heavy ductile iron castings can also be attributed to a better understanding of the solidification mechanics. In ductile iron, the graphite expansion pressure can be as high as 100 MPa locally, and if the mold wall yields, this expansion is lost, leading to porosity. We therefore paid great attention to mold compaction. The self-hardening sand molds were rammed with a vibration table to achieve a high and uniform mold hardness above 90 on the Bühlmann scale. This ensured that the mold would not deform during solidification, allowing the graphitic expansion to be utilized for self-feeding of the casting. We also kept the core gas evolution low by ensuring proper venting through the risers and core prints.
Another important aspect was the control of the chemical composition and inoculation of the ductile iron melt. The carbon equivalent was maintained at about 4.3% to promote a fully eutectic solidification, which minimizes the mushy zone width. The magnesium content was kept between 0.03% and 0.05% to ensure nodularity above 85%. We also performed a late inoculation with a barium-containing inoculant at a rate of 0.15% to prevent chill and to improve the graphite nucleation. These metallurgical measures, combined with the optimized casting design, contributed to the production of sound heavy ductile iron castings.
The use of computer simulation played an indispensable role in this optimization. We have used MAGMA for many years, and this case again demonstrated its value. The temperature field simulation allowed us to visualize the hot spots and to test different gating and risering configurations without costly shop trials. We also used the simulation to verify the effect of chills by varying their sizes and positions. The final process was validated both numerically and experimentally. A comparison of the simulated temperature gradients at a cut section of the casting before and after the modifications is presented in Table 3, showing that the temperature gradient in the critical zones increased significantly, favoring directional solidification.
| Location | Original Gradient (°C/cm) | Optimized Gradient (°C/cm) |
|---|---|---|
| Intake pipe support rib | 0.8 | 2.5 |
| Exhaust flange | 1.2 | 2.1 |
| Injector hole region | 0.9 | 1.9 |
| Oblique oil hole | 0.6 | 2.2 |
| Oil sump bottom | 1.0 | 1.8 |
To provide a more solid theoretical background, let us discuss the solidification of ductile iron in more detail. The eutectic solidification of ductile iron begins with the precipitation of austenite dendrites that grow ahead of the eutectic front. The graphite nodules grow in the liquid surrounded by austenite shells. As the temperature drops, the solubility of carbon in austenite decreases, causing carbon to diffuse onto the graphite nodules, which leads to an increase in the specific volume of the solid phase. This volume expansion is the primary driver of self-feeding in ductile iron castings. However, the expansion can also cause mold dilation, which in turn creates negative pressure in the interior, leading to the formation of shrinkage cavities. The modulus of the riser must be designed to keep the connection between the riser and the hot spot liquid until the solidification of the casting is complete. The critical feeding distance and the riser neck dimensions were calculated using the following empirical formula for ductile iron:
$$ D = 4.6 \sqrt[3]{V_c} $$
where \(D\) is the maximum feeding distance in inches and \(V_c\) is the volume of the feeding zone requiring a riser. For our cylinder head, we used the simulation to precisely determine the feeding zones, which were smaller than the theoretical maximum due to the complex geometry.
The introduction of the upper runner system was particularly beneficial. In the original bottom gating, the metal had to travel a long distance from the ingate to the top of the casting, losing superheat along the way. The upper runner allowed hot metal to enter directly into the upper section once the lower cavity was filled, maintaining the temperature of the intake pipe area above the liquidus. The flow velocity in the ingates was reduced from an average of 1.8 m/s to 0.8 m/s, which is below the critical value for dross formation in ductile iron. This reduction is important because dross from magnesium oxide and silicon oxide can form on the surface of the melt if the turbulence is excessive, and once entrained, it becomes a permanent inclusion. The open gating system with larger ingates also helped to keep the runner full, preventing air aspiration from the runner joints.
Another improvement was the placement of a riser on the oil sump core. Originally, this area had a high concentration of trapped gases because the core created a shallow pocket where gas bubbles accumulated. The riser, which extended to the top of the mold, provided an escape path for these gases. It also acted as a feeder for the oblique oil hole, which was located just below. The riser was designed with a neck diameter of 60 mm, and it was made with an insulating sleeve to slow down its solidification. The metal in the riser remained liquid for a period sufficiently long to feed the solidification contraction of the oil hole region.
In the case of the internal chill for the oblique oil hole, we used a mild steel rod with a diameter of 10 mm and a length of 150 mm, which was placed precisely in the center of the core. The rod was cleaned and coated with a thin layer of shellac to prevent rusting. During casting, the rod absorbed heat from the surrounding melt and increased the local cooling rate by a factor of about three. After machining, the rod is completely removed, leaving a clean oil passage. This approach is a classic example of using an internal chill in heavy ductile iron castings to eliminate isolated shrinkage, but it requires careful positioning to avoid the chill extending beyond the machined area.
We also experimented with the use of exothermic risers for the intake pipe top. An exothermic riser contains a chemical mixture that ignites on contact with the hot metal and releases additional heat, thereby extending the liquid lifetime of the riser. Although the cost is slightly higher than a conventional insulating riser, the improved yield and quality justified its use in this critical area. The exothermic riser was able to feed the support ribs of the intake pipe effectively, completely eliminating the cold shuts and shrinkage that had plagued the original process.
The results of the hydrostatic tests were particularly encouraging. After the process changes, the leakage rate dropped from 20% to less than 3%. The few leakages that still occurred were located in the exhaust flange area, where residual porosity was present but not severe. We subsequently added an external chill on that flange and increased the core vents to allow better gas escape. This further reduced the leakage to about 1%, and we are confident that with continued fine-tuning, this defect can be fully eliminated.
In conclusion, the optimization of the casting process for this heavy ductile iron cylinder head has been a comprehensive engineering effort involving defect analysis, computer simulation, gating and risering redesign, and the strategic use of chills and risers. The key lessons learned are:
- The gating system for heavy ductile iron castings must be designed to minimize turbulence and to avoid air entrapment. An open gating system with multiple ingates and an upper runner can achieve a smooth, quiescent filling pattern.
- The solidification shrinkage in ductile iron can be mitigated by using chills to create steep thermal gradients and risers to provide liquid metal feed. The combined action is more effective than either alone.
- Internal chills are a valuable tool for eliminating shrinkage defects in slender, machined passages.
- MAGMA simulation is an essential tool for guiding the process design and predicting defect formation in heavy ductile iron castings.
- Mold rigidity and proper compaction are crucial to utilize the graphitic expansion of ductile iron for self-feeding.
The scrap rate has been reduced from 70% to below 10%, and the quality of the castings is now consistently meeting the requirements. The methodology described here can be applied to other heavy ductile iron castings with similar challenges. We continue to monitor the process and to refine our simulation models to achieve further improvements in yield and reliability.
For future work, we plan to use simulation to optimize the exhaust flange area further, perhaps by modifying the core design or by adding a small blind riser. We also intend to explore the use of 3D-printed sand cores for the internal channels, which would allow more freedom in designing venting and chills. The successful production of this cylinder head has reinforced our conviction that a systematic approach, combining theoretical knowledge, simulation, and practical experience, is the key to producing high-quality heavy ductile iron castings.