In my daily work as a foundry engineer, I have frequently encountered the challenge of internal shrinkage in ductile iron casting. Shrinkage porosity is one of the most common and troublesome casting defects, and it becomes particularly severe when the material is ductile iron. Because most ductile iron compositions are eutectic or hypereutectic, the solidification mode is pasty or mushy. In a ductile iron casting, the outer shell does not develop the same mechanical resistance as gray iron, so the graphite expansion pressure during eutectic solidification can push the mold wall outward. This mold wall movement increases the overall volume of the casting and creates large internal voids. In addition, the last liquid to solidify between dendrites and eutectic cells cannot be fully fed by the remaining liquid metal, resulting in both macro- and micro-shrinkage. Therefore, the shrinkage tendency of ductile iron casting is higher than that of gray iron casting. When a customer reported internal shrinkage after machining a batch of cylinder heads, I had to investigate the root cause and implement corrective actions. In this article, I present my complete analysis, simulation results, process modifications, and verification data.
1. Product Parameters and Original Process
The component in question was a ductile iron cylinder head. The material specification was QT450-10, which corresponds to a ferritic ductile iron with a minimum tensile strength of 450 MPa and an elongation of 10%. The casting dimensions were 144 mm × 106 mm × 81 mm, and the single casting weight was 3 kg. The production line used a DISA vertical molding machine with a one-mold-four-cavity layout. The total pouring weight for the mold was 23 kg. The pouring temperature was controlled between 1360 °C and 1380 °C, and the pouring time was measured as 11.6 seconds. The original gating and risering system is summarized in the following table.
| Parameter | Original Value | Notes |
|---|---|---|
| Casting material | QT450-10 | Ferritic ductile iron |
| Overall dimensions | 144 mm × 106 mm × 81 mm | |
| Casting weight per piece | 3 kg | |
| Number of cavities | 4 | Vertical parting, two on top, two on bottom |
| Total mold weight | 23 kg | Including gating and risers |
| Pouring temperature | 1360–1380 °C | Measured in the ladle |
| Pouring time | 11.6 s | Manual pouring |
| Molding process | DISA vertical molding | High-pressure green sand mold |
In my analysis, the gating system was designed with a sprue, a runner bar, and ingates leading to four cavities. The risers were placed on top of the cylinder heads. One side used a spherical riser, and the other side used an oval riser. The original riser diameter was 80 mm. The pouring basin was relatively shallow, which later proved to be a critical factor.
2. Typical Defect and Root Cause Analysis
After the cylinder heads were delivered to the customer, a small batch machining operation revealed internal shrinkage inside the oil nozzle area of the upper cavities. Nondestructive testing was performed on 800 parts, and 41 parts showed internal shrinkage defects. The rejection rate due to shrinkage was 5.13%. This was unacceptably high. In my investigation, I used MAGMA simulation software to model the filling and solidification of the ductile iron casting. The simulation results are summarized in the following table.
| Simulation Result | Observation | Interpretation |
|---|---|---|
| Hot spot location | The oil nozzle region of the upper cavities showed a localized hot spot. | This region became the last to solidify, making it prone to shrinkage. |
| Shrinkage porosity prediction | The software predicted a large shrinkage defect in the same oil nozzle region. | Feeding was inadequate for that local thermal center. |
| Riser filling | The top risers were not completely filled in some molds. | Pouring basin depth was insufficient, causing short pouring. |
| Mold expansion | Graphitization expansion in ductile iron caused mild mold wall movement. | Increased the volumetric demand for liquid metal during solidification. |
After repeated casting trials and mold modifications, I categorized the root causes into three main areas. The first cause was related to pouring practice. To avoid iron spillage, the operator poured too shallowly into the pouring cup. Some molds had a runner bar and the top of the riser at the same level, which meant the riser could not be filled to its full height. I sectioned several castings with incompletely filled risers and confirmed the presence of shrinkage in those castings. The proportion of castings affected by the incompletely filled riser issue was about 21% of the sectioned samples, which corresponded to approximately 0.58% of total production.
The second cause was insufficient riser feeding capacity or restricted feeding path. After isolating the castings with obviously incomplete riser filling, I still found shrinkage in the remaining castings by nondestructive testing and sectioning. The bottom two cavities showed fewer shrinkage problems because gravity assisted the feeding of the bottom risers. This led me to conclude that the modulus of the top risers was too small for the local hot spot in the oil nozzle area.
The third cause was poor mold排气. The vertical mold had limited venting, which slowed down the liquid metal flow during pouring. In some cases, even after the pouring cup was filled, the mold continued to sink or settle, causing the riser to end up incompletely filled. The combination of insufficient pouring and poor venting made the ductile iron casting process unstable.
To quantify the feeding demand, I calculated the modulus of the hot spot and the riser. The modulus \(M\) is defined as the ratio of volume \(V\) to cooling surface area \(A\):
$$
M = \frac{V}{A}
$$
For a cylindrical riser with diameter \(D\) and height \(H\), the modulus can be expressed as:
$$
M_{\text{riser}} = \frac{\pi D^2 H / 4}{\pi D H + \pi D^2 / 4 \cdot 2}
$$
Assuming a riser height equal to its diameter, \(H = D\), the equation simplifies to:
$$
M_{\text{riser}} = \frac{D}{6}
$$
Therefore, for an 80 mm diameter riser, the modulus was:
$$
M_{80} = \frac{80}{6} \approx 13.3 \text{ mm}
$$
For a 86 mm diameter riser, the modulus became:
$$
M_{86} = \frac{86}{6} \approx 14.3 \text{ mm}
$$
The hot spot in the oil nozzle region required a feeding modulus larger than the original riser modulus. I estimated the hot spot modulus using the local geometry and found it needed at least 14 mm. This explained why the original riser was insufficient.
3. Process Adjustments and Implementation
3.1 Improved Pouring Practice
My first measure was to change the pouring procedure. Instead of a continuous pour that could be interrupted or stopped early, I implemented a delayed pouring technique. In this method, the pouring ladle is positioned over the pouring cup, and the molten iron is poured in a controlled manner with a slight delay to ensure that the cup remains full throughout the pour. This minimizes the risk of the riser being underfilled. I also instructed the operators to always fill the pouring cup completely and to avoid stopping the pour until the metal appeared in the riser vents.
The updated pouring parameters are shown below.
| Parameter | Before | After |
|---|---|---|
| Pouring method | Continuous, shallow cup | Delayed pouring, full cup |
| Pouring cup level | About 80% full | 100% full throughout the pour |
| Operator training | Minimal | Full instruction on riser filling recognition |
| Inspection after shakeout | Visual only, no riser check | All risers inspected, incomplete ones segregated |
In addition, I required the technicians to inspect the separated risers after shakeout. Any casting with an incompletely filled riser was immediately separated and rejected. This eliminated the obvious case of riser underfilling from the production stream.
3.2 Optimization of Riser Design
Since the top riser modulus was insufficient, I increased the riser diameter from 80 mm to 86 mm. The riser shape was changed as well. One side remained a spherical riser, and the other side became an oval or腰圆 riser. The riser neck or feeding通道 was also enlarged to improve feeding efficiency. I added a V-shaped groove around the riser neck to facilitate a better feeding path and to avoid premature freezing of the neck.
The design changes are summarized in the following table.
| Feature | Original Riser | Modified Riser |
|---|---|---|
| Diameter | 80 mm | 86 mm |
| Shape | Spherical on one side, oval on the other | Spherical on one side, oval on the other, but larger |
| Riser neck | Small, restricted | Enlarged with V-groove |
| Modulus | 13.3 mm | 14.3 mm |
| Feed volume | Sufficient for small hot spots only | Sufficient for the oil nozzle hot spot |
The feeding distance of a riser can be estimated using the thermal center distance. In ductile iron casting, the feeding distance \(L_f\) is related to the casting thickness \(t\) and the riser diameter \(D_r\). A conservative empirical equation for a plate-like section is:
$$
L_f = 2t + D_r
$$
For the oil nozzle region with an effective thickness of 12 mm and a riser diameter of 86 mm, the estimated feeding distance was:
$$
L_f = 2(12) + 86 = 110 \text{ mm}
$$
This was sufficient to cover the hot spot location, which was about 35 mm away from the riser neck. The original 80 mm riser gave:
$$
L_f = 24 + 80 = 104 \text{ mm}
$$
Though the difference seems small, the local geometry and the presence of the hot spot made the extra 6 mm in diameter critical. The improved modulus also reduced the solidification time of the riser relative to the casting. The solidification time \(t_s\) for a casting can be estimated by Chvorinov’s rule:
$$
t_s = B \left( \frac{V}{A} \right)^2
$$
where \(B\) is a molding material constant. Using the modulus squared, the increase from 13.3 mm to 14.3 mm gave a solidification time ratio of:
$$
\frac{t_{s,86}}{t_{s,80}} = \left( \frac{14.3}{13.3} \right)^2 \approx 1.156
$$
This means the modified riser stayed liquid about 15.6% longer than the original, allowing more time for gravity and graphite expansion to feed the hot spot.
3.3 Addition of Open Vent Riser
To address the poor mold排气 issue, I added an open vent hole at the top of each riser. This vent acted as an atmospheric vent and an escape path for gases. In a ductile iron casting, the mold gas permeability is often reduced by the fine sand and the mold coating. The open vent allowed the mold cavity to communicate with the atmosphere, which helped the riser to fill more completely and also improved the feeding behavior. The vent was positioned directly above the highest point of the riser, ensuring that gas pockets did not block liquid metal flow.
The effect of venting on the pressure inside the mold can be analyzed by the gas flow equation. The pressure drop \(\Delta P\) required to force gas through a vent of diameter \(d_v\) and length \(L_v\) is:
$$
\Delta P = \frac{32 \mu Q L_v}{\pi d_v^4}
$$
where \(\mu\) is the gas viscosity, \(Q\) is the volumetric gas flow rate, and \(d_v\) is the vent diameter. Increasing \(d_v\) significantly reduces the pressure drop because of the fourth power relationship. By adding a vent with an adequate diameter, I ensured that the pressure buildup during pouring was minimized, preventing the metal from flowing back or the mold from sinking.
| Parameter | Before | After |
|---|---|---|
| Mold venting | Only through parting line and core prints | Additional open vent at riser top |
| Vent diameter | None | 10 mm |
| Vent location | — | Riser top center |
| Effect | Slow filling, mold sinking | Fast and stable filling, no sinking |
I also re-ran the MAGMA simulation with the modified riser and venting system. The simulation predicted no shrinkage porosity in the oil nozzle region. The hot spot was completely eliminated because the larger riser and the improved feeding path allowed liquid metal to reach the hot spot until the end of solidification.
4. Verification and Production Results
After implementing all three changes, I carried out a small batch production trial. The trial consisted of 120 molds, corresponding to 480 cylinder heads. The castings were then inspected by the same nondestructive testing method used before. The inspection results are shown in the following table.
| Inspection Item | Before Modification | After Modification |
|---|---|---|
| Total parts inspected | 800 | 480 |
| Parts with internal shrinkage | 41 | 0 |
| Rejection rate | 5.13% | 0% |
| Customer machining complaints | Multiple | None |
| Riser underfilled cases | Approx. 0.58% | 0% |
In addition, I performed destructive sectioning on 10 trial castings from different mold positions. All sections were sound, with no visible shrinkage cavities or micro-porosity. The metallographic examination of the test pieces showed a typical ferritic structure with well-formed graphite nodules. The nodule count was in the range of 150 to 180 nodules per square millimeter, which is considered good for a ductile iron casting of this section size.
The final process parameters after optimization are summarized below.
| Process Parameter | Optimized Value |
|---|---|
| Molding process | DISA vertical molding, one mold four cavities |
| Pouring temperature | 1360–1380 °C |
| Pouring time | 11.6 s |
| Pouring method | Delayed pouring with full cup |
| Riser diameter | 86 mm |
| Riser type | Spherical on one side, oval on the other, with V-groove neck |
| Riser vent | Open vent 10 mm diameter at riser top |
| Simulation result | No shrinkage |
| Production verification | 0% shrinkage rejection |
5. Theoretical Considerations for Ductile Iron Casting Solidification
To fully understand why the modifications worked, I have considered the solidification of ductile iron casting in more detail. During eutectic solidification, austenite dendrites and graphite nodules form a coherent network. The remaining liquid becomes isolated in the interdendritic spaces. The graphite expansion can compensate for the solidification contraction, but only if the mold is rigid enough and the liquid can be fed back to the expanding regions. In green sand molds, mold wall movement may absorb some of the expansion, reducing the internal pressure. This is why the feeding system must be designed with adequate riser volume and modulus to ensure that no isolated liquid islands form.
The feeding efficiency \(\eta\) of a riser can be defined as:
$$
\eta = \frac{V_{\text{fed}}}{V_{\text{riser}}}
$$
where \(V_{\text{fed}}\) is the volume of liquid supplied to the casting, and \(V_{\text{riser}}\) is the total riser volume. For ductile iron, the required feeding volume is typically less than that for steel because the graphite expansion reduces the net contraction. However, the feeding path must remain open until the end of solidification. The enlarged riser neck and the V-groove kept the path open longer.
The cooling rate also affects the microstructure. The local solidification time \(\theta\) can be expressed as:
$$
\theta = \frac{M^2}{k}
$$
where \(k\) is a constant that depends on the mold material and the heat transfer coefficient. For a sand mold, \(k\) is relatively low, leading to long solidification times. With the original riser, the hot spot had a longer solidification time than the riser, so the riser froze first and could not feed. After the modification, the riser had a larger modulus, so its solidification time exceeded that of the hot spot, allowing feeding to continue.
| Quantity | Original Riser (80 mm) | Modified Riser (86 mm) |
|---|---|---|
| Volume (assuming \(H = D\)) | \(\frac{\pi 80^3}{4} \approx 402,124 \text{ mm}^3\) | \(\frac{\pi 86^3}{4} \approx 499,642 \text{ mm}^3\) |
| Modulus | 13.3 mm | 14.3 mm |
| Relative solidification time | 1.00 | 1.156 |
| Feeding distance (for 12 mm section) | 104 mm | 110 mm |
In my experience, these small numerical changes can be decisive in a ductile iron casting. The difference between a sound casting and a rejected casting is often a few millimeters in riser diameter or the presence of a vent. The combination of improved pouring, larger riser, and better venting provided a robust process window.
6. Lessons Learned and Best Practices
From this investigation, I have drawn several important lessons for the production of ductile iron casting components with complex geometries.
First, the design of the riser must always be checked by solidification simulation or hand calculations. The hot spot in the casting must have a smaller modulus than the riser. In this case, the hot spot was in the oil nozzle, which is a small but massive local region. The original riser did not provide enough modulus to feed it.
Second, the pouring practice must be controlled carefully. Even a perfect riser design will fail if the riser is not fully filled. The operator’s habit of pouring shallowly to avoid spillage was a major contributor. I corrected this by implementing a delayed pouring technique and by providing visual indicators on the mold so that the operator could confirm that the riser vents were filled.
Third, venting is essential in vertically parted molds, especially for ductile iron casting. The addition of an open vent at the riser top allowed the mold to breathe and prevented the metal from having to expel gas through the sand. This also reduced the back pressure that slowed down the filling.
Fourth, the quality of the ductile iron itself is important. The iron must have adequate carbon equivalent and low impurities to minimize shrinkage tendency. In this case, the chemistry was within specification, but the feeding system had to be adjusted to compensate for the natural shrinkage of ductile iron casting.
The carbon equivalent \(CE\) is calculated as:
$$
CE = C + \frac{Si}{4} + \frac{P}{2}
$$
For QT450-10, a typical composition is \(C = 3.7\%\), \(Si = 2.5\%\), and \(P < 0.05\%\). The carbon equivalent would be:
$$
CE = 3.7 + \frac{2.5}{4} + \frac{0.05}{2} = 3.7 + 0.625 + 0.025 = 4.35\%
$$
This high carbon equivalent gives a large amount of graphite expansion during eutectic solidification, which can help to reduce shrinkage if the mold is rigid. However, in a green sand mold, the expansion may be lost to mold wall movement. I considered the possibility of increasing mold hardness or using a less compressible molding sand, but the simplest solution was to improve the feeding system.
7. Conclusion
In conclusion, the internal shrinkage defects in the cylinder head were caused by a combination of underfilled risers, insufficient riser modulus, and poor mold venting. My systematic approach involved:
- Changing the pouring method to delayed pouring with a fully filled pouring cup.
- Increasing the riser diameter from 80 mm to 86 mm and enlarging the riser neck with a V-groove.
- Adding a 10 mm open vent at the top of each riser to improve exhaust and feeding.
These modifications were verified first by MAGMA simulation and then by production trials. The rejection rate due to shrinkage dropped from 5.13% to 0% in the trial batch. The final process has been running stably in mass production, and no customer complaints about internal shrinkage have been received since the changes.
Thus, for any ductile iron casting design, I always recommend careful attention to the following formula:
$$
M_{\text{riser}} > M_{\text{hot spot}}
$$
And the practical rule of always keeping the riser fully filled and well vented. By doing so, the risk of shrinkage can be almost eliminated, even for complex geometries such as cylinder heads.