As a quality engineer dedicated to automotive component manufacturing, I have spent a significant portion of my career addressing casting defects in ductile iron casting. The wheel hub, a critical rotating part of the vehicle chassis, demands excellent rigidity, fatigue resistance, and thermal conductivity. Our foundry specializes in producing ductile iron casting hubs and axle housings. These hubs are manufactured on a shared HWS high-pressure molding line with a green sand system. The mold box dimensions are \(1000 \times 800 \times 350/350\) mm, and each mold produces four hubs. The core is made of sodium silicate sand, and the gating system is top-poured. Melting is performed in an 8-ton medium-frequency induction furnace. Over a long production period, the most persistent and troublesome defect has been shrinkage porosity and macro-shrinkage, which primarily occur in the thick, heavy sections at the upper part of the hub.
My involvement in this project began when we noticed an unusually high rejection rate due to shrinkage cavities. The customer’s requirement for these ductile iron casting hubs is stringent, as any internal discontinuity can lead to premature fatigue failure under cyclic wheel loading. I therefore initiated a systematic investigation to identify the root causes and subsequently implement robust corrective actions. This paper describes my first-person experience and the technical measures we adopted to eliminate shrinkage porosity in this ductile iron casting part.
1. Root Cause Analysis of Shrinkage Porosity
After analyzing a large number of rejected hubs, I found that the shrinkage defects were consistently located in the heavy sections at the top of the casting, where the molten metal solidifies last. These regions act as hot spots because the thermal center remains partially liquid while the surrounding metal has already solidified. The resulting contraction of the liquid core cannot be compensated by liquid flow through the dendritic network, leading to porosity.
I first examined the chemical composition of the ductile iron casting. The measured values were as follows: carbon \(3.5\)–\(3.7\%\), silicon \(2.4\)–\(2.6\%\), manganese \( \le 0.5\%\), magnesium \(0.03\)–\(0.06\%\), sulfur \( \le 0.03\%\), phosphorus \( \le 0.07\%\), and rare earth \(0.01\)–\(0.05\%\). The graphite nodularity was rated as grade 1–4, and the nodule size was grade 6–7 according to the standard charts. The carbon equivalent was calculated using the formula:
$$CE = C + \frac{1}{3}(Si + P)$$
With the above ranges, the \(CE\) was approximately \(4.5\%\), which theoretically satisfies the requirement for a sound casting. However, I realized that the silicon content was at the lower end of the specification, which reduces its contribution to graphitization during eutectic solidification. In ductile iron casting, silicon promotes graphite precipitation and reduces the tendency for shrinkage porosity by increasing the volume expansion associated with graphite nodule growth. A lower silicon content therefore makes the casting more prone to shrinkage defects.
Another critical factor was the excessive amount of nodulizer used in the ladle treatment. We were adding \(1.6\)–\(2.0\%\) of nodulizer (FeSiMg8RE5) based on the weight of the molten metal. The residual magnesium content was consistently in the range of \(0.05\)–\(0.06\%\), which is very close to the upper control limit. High residual magnesium strongly increases the shrinkage tendency in ductile iron casting. The reason is that magnesium helps stabilize carbide and promotes undercooling, while also increasing the surface tension of the liquid. Both effects hinder the feeding of micro-shrinkage areas. Therefore, I concluded that the nodulizer addition rate was excessive for this specific hub geometry.
The pouring temperature was another parameter that I suspected. Our process specification called for a pouring temperature of \(1390\)–\(1400^\circ\mathrm{C}\). This is somewhat high for a ductile iron casting of this section size. Higher pouring temperatures increase the liquid contraction before solidification, and also extend the time for the hot spot to remain molten. Both effects increase the risk of shrinkage porosity. I decided to examine the possibility of lowering the pouring temperature while still ensuring complete mold filling.
Furthermore, the in-mold inoculation system on our automatic pouring machine was not functioning reliably. The flowmeter of the stream inoculation device could not accurately control the feed rate of the inoculant (FeSi75Al1.5). As a result, the amount of inoculant delivered into each mold varied significantly. This led to a situation where, within the same mold, some hubs exhibited shrinkage porosity while others were sound. When inoculation is insufficient, the number of heterogeneous nuclei and graphite nodules decreases, and the amount of non-spheroidal graphite increases. This increases intergranular segregation and promotes shrinkage porosity. Thus, inconsistent inoculation was a major contributor to the defect variability.
Finally, due to the product geometry and the constraints of the molding box, the top-poured gating system caused the hot molten metal to first impinge on the bottom of the cavity and then gradually fill upward. This created locally overheated regions in the upper part of the hub, which coincided with the thermal center where shrinkage occurred. The combination of a thick section, high local temperature, and inadequate feeding path made this area particularly vulnerable.
2. Development of Corrective Measures
Based on my root cause analysis, I formulated a multi-pronged approach to modify the ductile iron casting process. The goal was to reduce the shrinkage tendency by adjusting the chemical composition, optimizing the nodulization and inoculation practices, controlling the pouring temperature, and using external chills to accelerate solidification at the critical hot spot. Table 1 summarizes the process parameters before and after the improvements.
| Parameter | Before Improvement | After Improvement |
|---|---|---|
| Silicon content (%) | 2.4 – 2.6 | 2.6 – 2.8 |
| Nodulizer addition (%) | 1.6 – 2.0 | 1.4 – 1.6 |
| Residual Mg (%) | 0.05 – 0.06 | 0.04 – 0.05 |
| Rare earth content (%) | 0.01 – 0.05 | 0.02 – 0.04 |
| Pouring temperature (°C) | 1390 – 1400 | 1370 – 1380 |
| Inoculation control | Unreliable flowmeter | Precise feeding system |
| External chill | None | Three steel chills at hot spot |
The first modification was to increase the silicon content range from \(2.4\)–\(2.6\%\) to \(2.6\)–\(2.8\%\). I achieved this by adjusting the amount of ferrosilicon added during the ladle treatment and by introducing additional silicon through the late inoculation. This increase not only raised the carbon equivalent slightly but also improved the graphitization potential. The higher silicon content promotes the precipitation of carbon as graphite during the eutectic solidification, which produces a volumetric expansion that compensates for the solidification contraction of the metallic matrix. This is a well-known principle in ductile iron casting production. The silicon also increases the number of graphite nodules, which further reduces microshrinkage.
The second important change was to reduce the nodulizer addition from \(1.6\)–\(2.0\%\) to \(1.4\)–\(1.6\%\) of the melt weight. I also strictly controlled the rare earth content, adjusting it to \(0.02\)–\(0.04\%\). This lower nodulizer amount resulted in a residual magnesium content of approximately \(0.04\)–\(0.05\%\), which is still sufficient for adequate nodularity but does not overdraft the cast iron. The lower magnesium content reduces the shrinkage tendency because magnesium has a strong effect on the surface tension and on the amount of carbides formed during solidification. I verified the nodularity after this change and found that the graphite remained fully spheroidal, meeting the specification for the hub material.
The third measure involved replacing the unreliable stream inoculation flowmeter with a newly designed, precise feeding system. The modified system uses a gravimetric feeder with a closed-loop controller that maintains a constant flow rate of inoculant regardless of variations in the pouring stream. The set point for the inoculant flow rate was determined based on the pouring time and the desired amount of inoculant per mold. We found that a consistent addition of FeSi75Al1.5 at a rate of \(0.2\%\) of the mold weight produced the best results in terms of graphite nodule count and elimination of porosity. This control improvement was essential because the earlier variability in inoculation was causing severe inconsistency in the quality of the ductile iron casting.
I also redesigned the pouring practice. The pouring temperature was strictly controlled in the range of \(1370\)–\(1380^\circ\mathrm{C}\). This is lower than the original specification but still safe for complete mold filling, as verified by the surface quality of the castings. The lower temperature reduces the liquid contraction and shortens the solidification time, thereby decreasing the thermal gradient at the hot spot. The result was a significant reduction in the tendency for shrinkage porosity. To ensure consistency, I implemented a strict SOP for the furnace operators to check the temperature immediately before tapping and again just before pouring, with a tolerance of \(\pm 5^\circ\mathrm{C}\).
The most intriguing improvement, however, was the placement of external chills. The hot spot in this hub geometry is located at the root of the “petal” shaped flanges on the upper surface. Due to the mold box constraints, it was impossible to change the casting design or add risers in that area. Therefore, I chose to place three steel chills on the external surfaces corresponding to the hot spot. The chills were designed with specific dimensions to provide a high thermal diffusivity path, accelerating the cooling of the hot spot and thereby reducing the local solidification time. I used three different chills, designated as 1#, 2#, and 3#, because the curvature of the hub surface differed at each location. Table 2 lists the dimensions and placement of these chills.
| Chill designation | Dimensions (mm) | Placement description |
|---|---|---|
| 1# | 80 × 50 × 20 | On the top surface of the main hot spot area |
| 2# | 60 × 60 × 15 | On the side surface adjacent to the petal root |
| 3# | 100 × 40 × 25 | On the flange root facing the core cavity |
These chills were made of gray cast iron machined to fit the curvature of the pattern. They were coated with a thin layer of graphite wash to prevent sand adhesion and were placed manually into the mold cavity before closing. The chills acted as heat sinks, extracting heat from the hot spot at a much higher rate than the surrounding green sand. This accelerated the solidification of the thick section, allowing it to solidify earlier and thereby reducing the volume of the final liquid pocket that would otherwise shrink upon solidification. I was careful to ensure that the chills did not create any cold shuts or incomplete fusion at the surface; the resulting casting surface was subsequently machined, and the chills were not visible on the final product.
In addition to these five main measures, I also refined the carbon equivalent calculation to account for the higher silicon range. The new target was:
$$CE = C + \frac{1}{3}(Si + P) = 3.6 + \frac{1}{3}(2.7 + 0.05) \approx 4.52\%$$
While this value is similar to the previous \(CE\), the higher silicon content shifts the eutectic point and encourages a more favorable solidification pattern. The relationship between silicon content and shrinkage porosity can be expressed empirically by the following equation, which I derived from our production data:
$$V_{s} = k_1 \left(1 – \frac{Si}{2.5}\right) + k_2 (Mg – 0.04) + k_3 (T_p – 1380)$$
where \(V_s\) is a relative shrinkage index, \(k_1\), \(k_2\), and \(k_3\) are constants determined by regression, and \(T_p\) is the pouring temperature in degrees Celsius. This model helped me communicate the importance of each factor to the production team.
3. Implementation and Production Verification
After introducing the modified process, I conducted a small batch trial consisting of eight consecutive molds. Each mold contained four hubs, so a total of 32 hubs were produced and subsequently sectioned for internal examination. I personally oversaw the sectioning and dye penetrant inspection. The results were outstanding: none of the 32 hubs showed any signs of shrinkage porosity or macro-shrinkage. The internal soundness was further verified by ultrasonic testing on a sample of five hubs. The microstructure also improved: the graphite nodule count increased by approximately 20%, and the ferrite content was more uniform.
Table 3 provides a comparison of the defect rates before and after the implementation over a one-month production period.
| Period | Number of hubs inspected | Shrinkage porosity defects | Defect rate (%) |
|---|---|---|---|
| Before improvement (monthly average) | 4800 | 384 | 8.0 |
| After improvement (first month) | 5120 | 41 | 0.8 |
| After improvement (second month) | 5200 | 26 | 0.5 |
The reduction in defect rate from \(8\%\) to below \(1\%\) represents a significant financial benefit for our foundry, as well as improved delivery reliability to the customer. The scrap cost alone was reduced by approximately 90%, which more than justified the investment in the new inoculation control system and the chill fabrication.
I also measured the mechanical properties of the improved hubs to ensure they meet the Q450-10 specification. The tensile strength and elongation were evaluated on separately cast test bars as well as on samples taken from the hub flanges. Table 4 shows the average results:
| Property | Specification (Q450-10) | Average measured |
|---|---|---|
| Tensile strength (MPa) | ≥ 450 | 472 |
| Yield strength (MPa) | ≥ 310 | 338 |
| Elongation (%) | ≥ 10 | 12.5 |
| Brinell hardness (HB) | 160 – 210 | 187 |
The improved ductile iron casting also exhibited a more uniform hardness distribution across the thick and thin sections, which is an indirect confirmation of reduced segregation and porosity. The elimination of shrinkage defects improved the pressure tightness of the hubs, which is critical for applications where the hub is subjected to dynamic loading and potential crack initiation.
4. Discussion
My experience with this ductile iron casting project reinforced the importance of a holistic approach to defect elimination. Each of the five measures contributed to the overall success, and I believe it is essential to understand their combined effect. The silicon increase promoted graphitization; the reduced nodulizer lowered the residual magnesium; the precise inoculation increased the nodule count; the lower pouring temperature reduced liquid contraction; and the external chills accelerated solidification at the critical hot spot. All of these factors work together to minimize the time during which the casting has an unsupported liquid film, thereby preventing the formation of shrinkage porosity.
A useful theoretical framework for this behavior is the concept of “effective solidification interval” \(\Delta T_{eff}\). In ductile iron casting, the solidification interval is influenced by both the chemical composition and the cooling rate. The following empirical relationship can be written:
$$\Delta T_{eff} = a \cdot (Mg – 0.03) – b \cdot Si + c \cdot \ln\left(\frac{d}{d_0}\right)$$
where \(d\) is the local cooling rate, \(d_0\) is a reference cooling rate, and \(a,b,c\) are positive constants. The chills increase \(d\), thereby reducing \(\Delta T_{eff}\), while the higher Si and lower Mg also reduce \(\Delta T_{eff}\). The net result is a much shorter time for the dendrites to form and for the liquid to become isolated.
I also observed that the inoculation control improvement had a secondary benefit: the consistency of the microstructure improved dramatically. Before the change, some hubs showed a mixture of chunky and flake graphite, which are known to exacerbate shrinkage. After the change, all hubs exhibited uniform spheroidal graphite with a nodule count of over 150 per mm². This is a direct result of the stable supply of nucleation agents from the precise inoculant feeder.
The role of the external chills deserves special mention. In ductile iron casting, chills are often avoided because they can cause carbide precipitation if the cooling rate becomes too high. However, our chill design and placement were optimized to avoid this problem. The chill surface was small relative to the total hot spot, so the cooling rate remained within the range that promotes fine graphite but not carbides. We verified by metallographic examination that there were no carbides at the chill contacting surfaces. The thickness of the chill was chosen to absorb just enough heat to reduce the local solidification time by about 30%, which was sufficient to eliminate the porosity without creating hard spots.
Another critical insight from this project was the interaction between pouring temperature and mold filling time. We recorded the filling time for each mold and found that it increased slightly when we lowered the pouring temperature, but the flowability remained adequate. The lower temperature reduced the tendency for mold erosion and also decreased the amount of reoxidation dross, which further improved the casting soundness. This demonstrates that a small reduction in pouring temperature can have multiple positive effects on the ductile iron casting quality.
Based on my experience, I recommend that other foundries facing similar shrinkage defects in ductile iron casting should first perform a thorough statistical analysis of their process parameters. The key variables to examine are the residual magnesium, silicon content, inoculation consistency, and the local cooling rate at the defect site. The use of modeling software to simulate the solidification pattern can be highly beneficial in identifying the exact location of hot spots and in designing the appropriate chilling strategy.
In our case, the defect was always located at the same geometric feature, which made the chill solution particularly effective. If the defect had been distributed more randomly, a more comprehensive modification of the gating system would have been necessary. Nevertheless, the measures I implemented are broadly applicable to any ductile iron casting with a heavy top section. For instance, increasing the silicon content from \(2.4\%\) to \(2.7\%\) is generally safe as long as the ferrite content and elongation are not compromised. Similarly, reducing the nodulizer addition from \(1.8\%\) to \(1.5\%\) is feasible when the residual magnesium is closely monitored.
I also learned the importance of operator training and documentation. After the new inoculation control system was installed, I conducted several training sessions for the foundry operators to explain the relationship between inoculation and shrinkage defects. They became more vigilant in monitoring the flowmeter readings and in reporting any anomalies. The strict control of pouring temperature required better coordination between the melting shop and the molding line. I established a communication protocol whereby the pouring operator receives a signal from the furnace operator when the melt has reached the exact target temperature. This reduced the variance of the pouring temperature from \(\pm 15^\circ\mathrm{C}\) to \(\pm 5^\circ\mathrm{C}\).
The economic impact of these improvements was substantial. Prior to the changes, the scrap and rework cost for this hub was estimated at $12 per rejected casting. With a monthly production of about 4800 hubs and an 8% defect rate, this translated to a monthly loss of $4,608. After the improvements, the defect rate dropped to 0.8%, reducing the monthly loss to $460. This represents an annual savings of over $49,000, not including the reduction in downtime and the improved customer satisfaction. The investment in the new inoculation feeder and the chill fabrication was less than $20,000, so the payback period was less than five months.
I believe that the success of this project demonstrates the value of a disciplined, data-driven approach to solving casting defects. By carefully controlling the chemical composition, optimizing the treatment and inoculation processes, and using external chills at critical hot spots, we were able to produce sound ductile iron casting hubs with excellent mechanical properties. This experience has given me a deeper understanding of the complex solidification behavior of ductile iron and the many factors that can influence the formation of shrinkage porosity.
5. Conclusion
In conclusion, the shrinkage porosity in our ductile iron casting wheel hubs was successfully eliminated through a combination of five key modifications:
First, I increased the silicon content to \(2.6\)–\(2.8\%\) to enhance graphitization and raise the carbon equivalent effectively. Second, I reduced the nodulizer addition to \(1.4\)–\(1.6\%\) and controlled the rare earth range, thereby lowering the residual magnesium to a safer level. Third, I replaced the unreliable stream inoculation flowmeter with a precise gravimetric feeding system, which ensured consistent and adequate inoculation for every mold. Fourth, I placed three external steel chills at the hot spot location to accelerate cooling and reduce the local solidification time. Fifth, I strictly controlled the pouring temperature to \(1370\)–\(1380^\circ\mathrm{C}\).
The production verification with 32 sectioned hubs showed zero shrinkage defects, and the subsequent monthly defect rate dropped from \(8\%\) to less than \(1\%\). The mechanical properties remained well above the Q450-10 specification. The lessons learned from this project have been incorporated into our standard operating procedures for all ductile iron casting production. I am confident that these measures are sustainable and can be adapted to other similar casting geometries in our foundry. The continuous improvement mindset is essential in the foundry industry, and I am grateful for the opportunity to contribute to the reduction of casting defects and to the overall efficiency of our manufacturing process.
For any foundry dealing with shrinkage porosity in ductile iron casting, I strongly recommend starting with a thorough examination of the chemistry and inoculation practices. Often, the root cause is a combination of several factors that individually appear to be within specification, but collectively increase the shrinkage tendency. By using statistical tools and controlled experiments, it is possible to identify the optimal process window. External chills are a powerful and inexpensive solution when the defect location is predictable. However, they must be designed with care to avoid introducing new defects such as carbides or gas entrapment.
This project has been a rewarding experience in my engineering career. I have learned that solving a complex casting defect requires not only technical knowledge but also effective communication and teamwork with the production staff, maintenance team, and quality control personnel. The successful implementation of these improvements has strengthened our reputation as a reliable supplier of high-quality ductile iron casting components. I hope that sharing my practical experience can help other engineers facing similar challenges in their own foundries.
Finally, I would like to emphasize that the principles discussed here are not limited to wheel hubs. Any steel or iron casting that has a thick section acting as a hot spot can benefit from the combined application of composition adjustment, controlled inoculation, lower pouring temperature, and chills. The key is to understand the solidification kinetics and to use all available tools to promote directional solidification and adequate feeding. With careful engineering and rigorous process control, defects like shrinkage porosity can be virtually eliminated, leading to higher yields and improved product performance in ductile iron casting.