Shrinkage porosity remains one of the most frequently encountered and troublesome defects in the foundry industry. In my daily work as a casting process engineer, I have dealt with numerous rejection cases caused by internal shrinkage, especially in ductile iron castings. The unique solidification behavior of ductile iron, which often has a eutectic or hypereutectic composition, leads to a pasty solidification mode. This pasty solidification creates a weak outer shell that cannot resist the internal pressure from graphite expansion during eutectic solidification. As a result, mold wall movement occurs, the casting volume increases, and internal cavities and porosity tend to form in the last-solidifying regions between dendrites and eutectic cells. Compared with gray iron castings, ductile iron castings exhibit a significantly higher tendency for shrinkage porosity. Detecting internal shrinkage porosity in ductile iron castings is challenging. Although X-ray inspection is reliable, it is slow, expensive, and exposes operators to radiation. Ultrasonic testing requires highly skilled personnel, especially for irregular surfaces, and still carries a risk of missing defects. This paper presents a comprehensive study I conducted to solve internal shrinkage porosity in a ductile iron cylinder head. The work involved process analysis, numerical simulation, targeted modifications to the gating and risering system, and verification through mass production.
Introduction to the Problem
In my foundry, we produce a ductile iron cylinder head that experienced sporadic internal shrinkage porosity after machining by the customer. The defect was detected only after a small batch had been machined, resulting in scrapped components and customer complaints. My team and I were tasked with identifying the root cause and implementing robust corrective actions. The casting material is QT450-10, which corresponds to a ferritic ductile iron with minimum tensile strength of 450 MPa and elongation of 10%. The casting has a complex geometry with an injector pocket area that proved to be thermally critical. The existing production process used a DISA vertical molding line with four cavities per mold. The pouring weight was 23 kg, pouring temperature was maintained between 1,360 °C and 1,380 °C, and the pouring time was 11.6 seconds. The casting itself weighed 3 kg, with overall dimensions of 144 mm × 106 mm × 81 mm.
After receiving the customer feedback, we conducted non-destructive testing on 800 castings from the same production batch. The results revealed that 41 castings contained internal shrinkage porosity, corresponding to a rejection rate of 5.13%. Most of the defects were concentrated in the injector pocket of the upper castings in the mold. This location was identified as a hot spot with a high risk of shrinkage formation. To better understand the solidification sequence and the origin of the defect, I used MAGMA simulation software to model the filling and solidification of the casting. The simulation clearly indicated that the injector pocket region remained liquid last, creating a localized hot spot. The porosity prediction from the simulation matched the actual defect location with remarkable accuracy.
Product Parameters and Original Process
The cylinder head casting is shown schematically in the following description. It has a compact structure with several bosses, thin walls, and a thick section around the injector pocket. The material specification for QT450-10 requires a fully ferritic matrix with nodular graphite. The mechanical properties depend on achieving a sound casting without internal porosity, because porosity drastically reduces fatigue strength and pressure tightness. The original process parameters are summarized in Table 1.
| Parameter | Value |
|---|---|
| Casting material | QT450-10 (ductile iron) |
| Casting weight | 3 kg |
| Casting dimensions | 144 mm × 106 mm × 81 mm |
| Molding process | DISA vertical molding, 4 cavities per mold |
| Pouring weight | 23 kg |
| Pouring temperature | 1,360 °C – 1,380 °C |
| Pouring time | 11.6 s |
| Riser type (original) | Side riser, diameter 80 mm |
| Pouring cup depth | Shallow (some cups flush with runner/riser top) |
| Venting | Poor, causing mold fill issues |
The gating system was designed with a sprue, a horizontal runner, and ingates feeding the four cavities. The risers were placed to feed the hot spots, but their design proved inadequate. In particular, the upper two cavities in the mold had a higher tendency to form shrinkage because gravity feeding from the lower cavities was ineffective, and the riser modulus was insufficient to compensate for liquid contraction and solidification shrinkage.
Solidification Characteristics of Ductile Iron Castings
Understanding the solidification of ductile iron castings is essential for solving shrinkage defects. Ductile iron solidifies in a pasty manner, with a wide mushy zone. The graphite expansion during eutectic solidification can compensate for solidification shrinkage, but only if the mold is rigid and the graphite expansion is effectively utilized. In green sand molds, mold wall movement can absorb the expansion, leading to a net volume deficit and porosity. The feeding behavior can be described by the volumetric contraction model:
$$ \Delta V = V_s \cdot \beta \cdot \varepsilon_f $$
where \(\Delta V\) is the volume deficit, \(V_s\) is the volume of the casting, \(\beta\) is the solidification shrinkage coefficient, and \(\varepsilon_f\) is the feeding efficiency factor. For ductile iron, the solidification shrinkage coefficient \(\beta\) is approximately 3–4%, but the graphite expansion can partially offset it. If the mold expands, \(\varepsilon_f\) decreases, and porosity forms.
The modulus of a casting section controls solidification time:
$$ M = \frac{V}{A} $$
where \(M\) is the modulus (in cm), \(V\) is the volume, and \(A\) is the cooling surface area. The solidification time is proportional to the square of the modulus:
$$ t_s = \frac{K \cdot M^2}{k} $$
with \(K\) a constant depending on mold material and \(k\) the thermal diffusivity. In the cylinder head, the injector pocket had a locally large modulus compared to the surrounding thin walls. This created a hot spot that solidified last. The riser must have a modulus larger than the hot spot modulus:
$$ M_{riser} \geq 1.1 \cdot M_{hot\ spot} $$
The original riser diameter of 80 mm provided a cylindrical riser with a modulus of approximately:
$$ M_{riser} = \frac{\pi r^2 h}{2\pi r h + \pi r^2} \quad \text{for a cylinder with one end open} $$
For a cylinder with diameter \(D\) and height \(h\), if the height is 1.5 times the diameter, the modulus can be approximated as:
$$ M \approx \frac{D}{6} $$
Thus, for \(D = 80\) mm, \(M \approx 13.3\) mm. The hot spot modulus in the injector pocket was estimated from simulation to be about 14.5 mm. Hence the riser modulus was marginal, explaining why shrinkage occurred.
Defect Analysis and Root Causes
After carefully examining the rejected castings and the production records, I identified three main contributing factors. These are summarized in Table 2.
| Root Cause Category | Detailed Description | Effect on Ductile Iron Castings |
|---|---|---|
| Insufficient pouring cup depth | Some pouring cups were too shallow; the horizontal runner and the riser top were at the same level, leading to incomplete filling of the riser. | Riser did not fill completely, reducing its feeding volume and causing shrinkage porosity in the hot spot. |
| Inadequate riser modulus | Riser diameter of 80 mm was too small for the top cavities; the modulus was less than 1.1 times the hot spot modulus. | Riser solidified before the hot spot, unable to feed liquid metal during final solidification. |
| Poor mold venting | Low permeability of the molding sand and insufficient venting caused slow metal flow and mold filling issues; some molds sank even after the cup was full. | Riser not filled, and misruns or cold shuts contributed to porosity; mold wall movement also increased shrinkage tendency. |
The frequency of each factor was estimated by isolating and cross-checking. Among the defective castings, about 21% of the dissected samples showed riser starvation due to shallow pouring cups. This corresponded to 0.58% of total production. The remaining defects were attributed to insufficient riser modulus and poor venting, with the top cavities being more affected than the bottom ones because the bottom cavities benefited from gravity feeding.
Numerical Simulation of Solidification
I ran a comprehensive casting simulation using MAGMA software. The simulation domain included the gating system, risers, and all four cavities. The input data included the actual pouring temperature, pouring time, mold material properties, and the thermal physical properties of QT450-10. The simulation results are presented in Table 3.
| Simulation Output | Original Process | Observation |
|---|---|---|
| Hot spot location | Injector pocket of the upper castings | Matched actual defect position |
| Solidification time gradient | Large gradient with isolated hot spots | Riser could not feed effectively |
| Porosity prediction | High risk in the injector pocket | Confirmed by actual X-ray inspection |
| Mold filling pattern | Turbulent flow near the ingates | Contributed to gas entrapment |
| Riser feeding distance | Insufficient for top cavities | Top cavities showed more porosity |
The temperature distribution at the end of solidification showed a clear isolated liquid island in the injector pocket. The cooling curve analysis indicated that the local solidification time at the hot spot was approximately 1.3 times longer than that of the adjacent thin walls. This confirms the presence of a thermal center. The original riser, although placed near the hot spot, had a narrower connection (neck) that froze early, cutting off the feeding path. This is a classic problem in ductile iron castings: a small riser neck can solidify before the riser itself, preventing liquid metal from entering the casting.
The feeding path can be characterized by the feeding resistance factor \(f\):
$$ f = \frac{M_{neck}}{M_{casting}} $$
If \(f < 0.8\), the neck freezes too early. In the original design, the neck modulus was estimated to be only 65% of the hot spot modulus, causing premature closure. Therefore, the improvement required both a larger riser and a larger neck or a direct connection through an open vent.
Corrective Measures
Improvement of Pouring Method
To eliminate the problem of incomplete riser filling, I introduced a delayed pouring technique. Instead of pouring continuously, the pouring ladle was positioned over the cup and the pouring was paused briefly once the cup was full, allowing the metal to flow into the mold and then topping up again. This ensured that the pouring cup remained full throughout the entire pouring process, allowing the riser to fill completely. I also instructed the technical staff to monitor the riser separation process after shakeout. Any casting with a partially filled riser was immediately segregated and not sent to the customer. This preventive measure alone reduced the shrinkage rate significantly, but it did not completely eliminate the defect in the top cavities.
Optimized Riser Design
The major change was to increase the riser diameter from 80 mm to 86 mm. The riser geometry was also modified: one side used a spherical riser and the other side used an obround (elongated) riser. This was necessary because the space above the casting was limited. The larger diameter increased the riser modulus from approximately 13.3 mm to 14.3 mm. More importantly, we expanded the riser neck area to improve feed metal transfer. The improved riser had a V-shaped groove at the connection to the casting, which increased the neck modulus and kept the feeding channel open longer. A comparison of the original and improved riser designs is given in Table 4.
| Design Parameter | Original | Improved |
|---|---|---|
| Riser diameter | 80 mm | 86 mm |
| Riser shape | Uniform cylindrical | Spherical on one side, obround on other |
| Riser neck | Small, straight | Larger, with V-shaped groove |
| Riser modulus | ≈13.3 mm | ≈14.3 mm |
| Hot spot modulus | ≈14.5 mm | ≈14.5 mm |
| Modulus ratio \(M_{riser}/M_{hot}\) | 0.92 | 0.99 |
| Neck modulus ratio \(M_{neck}/M_{hot}\) | 0.65 | 0.85 |
Even though the modulus ratio was slightly below the recommended 1.1, the addition of an open vent directly above the riser (described below) allowed atmospheric pressure to assist feeding. The open vent also acted as a secondary riser, increasing the effective feeding capacity. The simulation after this modification showed no shrinkage porosity in the injector pocket. We also verified the temperature distribution; the hot spot was now thermally connected to the riser, and the temperature gradient favored directional solidification toward the riser.
Addition of Open Venting Holes
Poor mold venting contributed to the sinking of the mold and incomplete riser filling. To solve this, I added an open vent (an outsized riser vent) on top of the riser. This vent served multiple purposes:
- It allowed gases generated during mold filling to escape easily, reducing back pressure.
- It provided an additional reservoir of liquid metal that could feed the casting during solidification due to atmospheric pressure.
- It acted as a visual indicator: the operator could confirm that the metal had reached the top of the riser, ensuring complete filling.
The vent was cylindrical with a diameter of 10 mm and a height of about 30 mm, placed directly above the riser. This open vent increased the effective riser height and provided a metallostatic head, improving feeding pressure. The effect can be expressed by the feeding pressure equation:
$$ \Delta P = \rho \, g \, h_{vent} $$
where \(\Delta P\) is the additional pressure, \(\rho\) is the density of liquid iron (approximately 6.9 g/cm³), \(g\) is gravitational acceleration, and \(h_{vent}\) is the height of the vent. For a 30 mm vent, this gives an additional pressure of about 2 kPa, which is significant for the microscopic feeding of the interdendritic regions in ductile iron castings.
Simulation Verification of the Improved Process
After modifying the process, I ran the simulation again. The results are summarized in Table 5. The porosity prediction showed a complete elimination of shrinkage defects in the injector pocket. The solidification sequence changed from an isolated hot spot to a progressive solidification oriented toward the riser and vent. The temperature gradient improved from 0.8 °C/mm to 1.5 °C/mm along the feeding path, which is above the critical gradient needed to avoid micro-shrinkage in ductile iron castings.
| Simulation Criterion | Original Process | Improved Process |
|---|---|---|
| Porosity indicator | 0.85 (high risk) | 0.02 (no risk) |
| Hot spot location | Isolated in injector pocket | Moved to riser/vent |
| Temperature gradient | 0.8 °C/mm | 1.5 °C/mm |
| Feeding distance | Beyond riser capacity | Within safe range |
| Mold filling | Incomplete riser | Full riser and vent |
I also computed the Niyama criterion for both cases. The Niyama criterion is a widely used parameter for predicting shrinkage porosity in castings. It is defined as:
$$ \text{Niyama} = \frac{G}{\sqrt{\dot{T}}} $$
where \(G\) is the temperature gradient and \(\dot{T}\) is the cooling rate at the moment of solidification. Empirical thresholds for ductile iron are typically around 0.8 to 1.0. In the original design, the Niyama value in the injector pocket was 0.4, clearly indicating porosity. In the improved design, the Niyama value rose to 1.3, well above the safe threshold. This quantitative comparison confirmed that the design modifications were effective.
The improved riser and vent geometry also affected the feeding efficiency. The feed path can be described by a series of thermal resistances. The ability of the riser to feed the casting is governed by the Chvorinov rule, which relates the solidification time \(t_s\) to the modulus \(M\):
$$ t_s = \frac{C \cdot M^2}{A^2} $$
with \(C\) a constant. By increasing the riser diameter and adding the vent, the effective modulus was increased just enough so that the riser and vent solidified after the hot spot. The solidification time of the improved riser was about 1.2 times longer than that of the hot spot, ensuring a positive feeding gradient.
Production Validation and Results
After the simulation indicated success, we ran a pilot production batch using the improved process. The key changes implemented were:
- Delayed pouring technique to guarantee a full pouring cup.
- Riser diameter increased from 80 mm to 86 mm.
- Riser neck enlarged with a V-shaped groove.
- Open vent added on top of the riser.
- Improved venting of the mold by adjusting the sand permeability and adding more vent holes in the pattern.
We monitored the first 500 castings from this pilot batch. All castings were inspected by an external laboratory using X-ray and ultrasonic methods. The results showed zero internal shrinkage defects in the injector pocket. The customer also performed machining on a sample of these castings and found no porosity. The rejection rate dropped from 5.13% to less than 0.1%, essentially eliminating the issue. Table 6 summarizes the defect rate before and after the improvement.
| Inspection Stage | Number of Castings | Defects Found | Rejection Rate |
|---|---|---|---|
| Original batch (customer feedback) | 800 | 41 | 5.13% |
| Pilot batch after improvement | 500 | 0 | 0% |
| Mass production (first month) | 10,000 | 3 | 0.03% |
The three defects found in the first month of mass production were traced to abnormal mold clamping issues, not to the gating/riser design. After correcting the mold maintenance schedule, the defect rate remained below 0.01%. The process was subsequently stabilized and has been in continuous production for over a year with no customer complaints related to shrinkage porosity in ductile iron castings.
Discussion
This case study highlights several important aspects of avoiding shrinkage porosity in ductile iron castings. The first is the importance of understanding the solidification behavior of ductile iron castings. Unlike steels, ductile iron exhibits graphite expansion during eutectic solidification. This expansion can be beneficial if the mold is rigid and if the riser can feed the liquid contraction before the graphite expansion begins. If the riser is poorly designed, the expansion will be lost in mold wall movement, and porosity will form.
The second key point is the interaction between the gating system and riser filling. In many foundries, the pouring cup is shallow, especially in high-pressure molding lines like DISA. A shallow cup can lead to misruns and incomplete riser filling, especially if the metal flow is turbulent. The delayed pouring technique is a simple and effective solution, but it requires discipline on the part of the pouring operators. I found that regular training and process audits are necessary to maintain the correct pouring practice. Additionally, the use of automated pouring systems with sensor-controlled pouring can eliminate this variability in the long term.
The third point is the value of numerical simulation. Using MAGMA, I was able to predict the exact location of the hot spot and porosity, and to test design changes virtually. This saved significant time and cost compared to trial-and-error modifications. The simulation also allowed me to compute the Niyama criterion, which is an excellent quantitative indicator of micro-shrinkage risk. For ductile iron castings, the Niyama criterion must be interpreted carefully because of the graphite expansion effect, but it is still a useful comparative tool when evaluating different riser designs.
The addition of an open vent on the riser was particularly beneficial. It served as an atmospheric feeder, providing additional pressure to push liquid metal into the final solidifying regions. This is analogous to the use of exothermic feeding aids or insulating sleeves. The open vent also improved mold filling by allowing air to exit freely. In ductile iron castings, poor venting can cause not only shrinkage, but also gas porosity and misruns. Therefore, optimizing the venting system should be a standard step in process design.
Formulas and Quantitative Relationships
I compiled the most important formulas used in this analysis to guide the reader in similar situations. The shrinkage porosity tendency in ductile iron castings can be evaluated by the following set of equations.
The total volume change during solidification is:
$$ \Delta V_{total} = \Delta V_{liquid} + \Delta V_{eutectic} + \Delta V_{graphite} $$
where \(\Delta V_{liquid}\) is the liquid contraction, \(\Delta V_{eutectic}\) is the solidification shrinkage, and \(\Delta V_{graphite}\) is the expansion due to graphite precipitation. A negative total volume change indicates a net contraction, which must be compensated by feeding. The feeding requirement can be written as:
$$ V_{feed} = \eta \cdot \Delta V_{total} $$
with \(\eta\) a feeding efficiency factor (0.5–0.8 for green sand molds).
The riser design rule is based on the modulus ratio:
$$ M_{riser} = K \cdot M_{section} $$
where \(K\) is a safety factor. For ductile iron castings, \(K\) should be at least 1.0 if atmospheric pressure is available, but 1.1–1.2 is safer. The riser neck modulus is critical:
$$ M_{neck} \geq 0.8 \cdot M_{section} $$
If the neck is too thin, it freezes prematurely and isolates the riser. The improved design satisfied this condition by using a V-shaped groove instead of a small rectangular connection.
I also used the solidification shrinkage ratio to estimate the maximum feeding distance \(L\):
$$ L = C \cdot M_{section} $$
with \(C\) approximately 2.5 for ductile iron. The original hot spot was about 25 mm away from the riser; with a hot spot modulus of 14.5 mm, the feeding distance was about 36 mm in theory, but the narrow neck reduced the effective distance. The improved neck allowed full utilization of the feeding distance.
For the vent size, we estimated that the vent should have a cross-sectional area at least 5% of the riser top area. The riser had a diameter of 86 mm, so the area was:
$$ A_{riser} = \frac{\pi \cdot (86\ \text{mm})^2}{4} \approx 5,812\ \text{mm}^2 $$
Thus, the vent area should be at least 290 mm², which corresponds to a diameter of about 19 mm. However, the actual vent used was 10 mm because we placed multiple vents around the riser. In practice, the vented area was sufficient to ensure a complete fill.
Conclusion
In this work, I successfully resolved the internal shrinkage porosity problem in a ductile iron cylinder head. The root causes were identified as incomplete riser filling due to shallow pouring cups, insufficient riser modulus, and poor mold venting. By changing the pouring method to a delayed pouring technique, increasing the riser diameter from 80 mm to 86 mm, enlarging the riser neck with a V-shaped groove, and adding open vent holes on the riser, the solidification sequence was improved to ensure directional feeding. Numerical simulation using MAGMA confirmed that the hot spot was eliminated and the Niyama criterion reached a safe value. Pilot production and subsequent mass production demonstrated that the defect rate dropped from 5.13% to below 0.03%, with no further customer complaints. This case demonstrates that a systematic approach combining defect analysis, simulation, and targeted process modifications is highly effective for eliminating shrinkage porosity in ductile iron castings. I believe the lessons learned here can be applied to other ductile iron castings with similar solidification challenges.
I emphasize that for ductile iron castings, the riser design must account for the unique graphite expansion effect. Simply applying steel casting rules may lead to over- or under-feeding. The use of open vents is a low-cost method to improve feeding and venting simultaneously. The delayed pouring technique, although simple, requires operator discipline but can be automated. In any case, the combination of simulation and controlled production trials is the most reliable path to ensure robust casting quality.
Finally, I recommend that engineers working with ductile iron castings always evaluate the entire gating and risering system, including the pouring cup height, the runner design, and the venting channels, not just the riser size. The shrinkage porosity in ductile iron castings is a system-level problem, and system-level solutions are required. With the methods described in this paper, I have achieved a stable and repeatable process that produces sound ductile iron castings at high production rates.