Ductile iron castings are widely used in commercial vehicles because they combine high strength, good machinability, and excellent damping properties. In particular, wheel hubs are safety-related rotating components. They must possess adequate rigidity, fatigue resistance, and heat dissipation. At the foundry where I work, ductile iron castings are produced on a high-pressure green-sand molding line. Among the most important ductile iron castings made there are wheel hubs and axle housings. The hub is a key product and is manufactured together with other castings on the same line.
I have spent considerable time trying to understand why these ductile iron castings exhibited shrinkage porosity in the upper heavy sections. The original composition was approximately 3.5–3.7% C, 2.4–2.6% Si, Mn up to 0.5%, Mg from 0.03% to 0.06%, S up to 0.03%, P up to 0.07%, and rare earth from 0.01% to 0.05%. Although the carbon equivalent was about 4.5%, the shrinkage defects were not eliminated. In this article I describe the full diagnostic procedure, the root causes I identified, and the process changes that finally solved the problem.
Initial Production Route
The production route for these ductile iron castings was already well established. The material specification was a ferritic grade requiring a tensile strength of approximately 450 MPa and an elongation of about 10%. Each casting weighed about 46 kg. The molding system was an HWS static-pressure green-sand line. The flask size was 1000 mm by 800 mm with cope and drag heights of 350 mm each. Every mold produced four castings. A sodium-silicate-bonded sand core was used to form the internal shape, and the gating system was a top-pouring arrangement. Melting was carried out in an 8-tonne medium-frequency induction furnace. Nodularization was performed in the treatment ladle using a FeSiMg8RE5 nodulizer. The inoculant used during pouring was FeSi75Al1.5.
| Parameter | Original Value |
|---|---|
| Casting material | Ductile iron grade 450-10 |
| Casting mass | Approximately 46 kg |
| Molding process | HWS static-pressure green-sand molding |
| Flask size | 1000 mm x 800 mm x 350 mm / 350 mm |
| Cavities per mold | 4 |
| Core material | Sodium-silicate-bonded sand |
| Gating system | Top pouring |
| Melting unit | 8-tonne medium-frequency induction furnace |
| Pouring temperature | 1390–1400 °C |
| Observed defect | Shrinkage porosity in thick upper regions |
Defect Morphology and Location
In the first stage of the investigation, I collected defective castings from the shake-out area. The shrinkage defects were not uniformly distributed. They appeared repeatedly in the thick upper sections of the wheel hub, especially in the heavy pad-like areas and at the roots of the petal-shaped flanges. Some defects were visible on the treated surface, while others were hidden internally and only observed after machining or sectioning.
I classified the defects as shrinkage cavities and shrinkage porosity. The cavities had rough, oxidized internal surfaces when they were open to the casting surface. Internal porosity appeared as branched networks of small voids concentrated near the final solidification zones. This is a common problem in ductile iron castings when the feeding path to the last liquid becomes blocked before solidification is complete.
Carbon Equivalent and Composition Analysis
The carbon equivalent calculation is a standard first step for ductile iron castings. I used the conventional expression:
\[
\mathrm{CE} = C + \frac{\mathrm{Si} + \mathrm{P}}{3}
\]
For the upper limits of the original composition, the carbon equivalent reached about 4.5%. This value is often considered sufficient for producing sound ductile iron castings. However, the silicon content was low in relation to the carbon content. Silicon is an important graphitizing element. A lower silicon concentration reduces the amount of graphite precipitation during eutectic solidification, and therefore reduces the internal expansion that helps feed shrinkage in ductile iron castings.
The contribution of a small silicon change to the carbon equivalent is:
\[
\Delta \mathrm{CE} \approx \frac{\Delta \mathrm{Si}}{3}
\]
If I increased silicon by 0.2%, the carbon equivalent would increase by only about 0.07%. That seems small. However, the real benefit is not only in the numerical value of CE. Silicon strongly influences graphite nucleation, carbon diffusion, and the timing of graphite expansion. In ductile iron castings, graphite expansion must occur while the solid network is still able to push liquid into the shrinking intergranular spaces. If silicon is too low, graphite forms later and the expansion is less effective.
| Element | Original Measured Range |
|---|---|
| C | 3.5–3.7% |
| Si | 2.4–2.6% |
| Mn | ≤ 0.5% |
| S | ≤ 0.03% |
| P | ≤ 0.07% |
| Mg | 0.03–0.06% |
| Rare earth | 0.01–0.05% |
| CE | About 4.5% at the upper end |
Solidification and Shrinkage Mechanisms
To understand why shrinkage appeared in these ductile iron castings, I considered the volumetric changes that occur during cooling and solidification. A simplified volume balance can be written as:
\[
\Delta V_{\mathrm{net}} = \Delta V_{\mathrm{liquid}} + \Delta V_{\mathrm{solidification}} – \Delta V_{\mathrm{graphite}} – \Delta V_{\mathrm{feeding}}
\]
Here, \(\Delta V_{\mathrm{liquid}}\) is the contraction of liquid iron during cooling from the pouring temperature to the liquidus temperature. \(\Delta V_{\mathrm{solidification}}\) is the contraction associated with the austenite-liquid transformation. \(\Delta V_{\mathrm{graphite}}\) is the expansion caused by graphite precipitation. \(\Delta V_{\mathrm{feeding}}\) is the volume supplied by risers, feeders, or external liquid. If the net value remains negative at the final stage of solidification, shrinkage porosity forms.
For ductile iron castings, graphite expansion is especially important. Graphite has a low density compared with liquid iron. When graphite nodules precipitate, the solid-liquid mixture increases in volume. This expansion can offset the previous liquid contraction. The challenge is that graphite expansion must be used before the mold cavity becomes completely isolated by solidified metal. If the local thermal gradient is low, the last liquid remains trapped in the center of a thick section. That trapped liquid contracts and creates porosity.
A useful criterion for evaluating shrinkage risk is the Niyama criterion:
\[
N_y = \frac{G}{\sqrt{\dot{T}}}
\]
where \(G\) is the temperature gradient in the liquid at the end of solidification, and \(\dot{T}\) is the local cooling rate. A low Niyama value indicates a high probability of microporosity. In the defective ductile iron castings, the heavy top section had a small gradient and slow cooling. The Niyama criterion was therefore low in exactly the region where the shrinkage defects were observed.
Root-Cause Diagnosis
After the initial composition and solidification analysis, I identified five interacting causes. They are summarized below.
| Cause | Effect on Ductile Iron Castings |
|---|---|
| Low silicon content | Reduced graphitization, weaker graphite expansion, larger shrinkage tendency |
| Excessive nodulizer addition | High residual magnesium, higher undercooling, more inclusions, increased shrinkage |
| High pouring temperature | More liquid contraction, more superheat to be removed, heavier feeding demand |
| Unstable stream inoculation | Fewer graphite nuclei, non-spherical graphite, intergranular shrinkage |
| Thermal hot spots | Final liquid isolated in thick sections, no important feeding path, porosity in that volume |
Low Silicon Content
The first root cause was the silicon range. Although the carbon equivalent was close to 4.5%, the silicon range of 2.4–2.6% was too low for this casting geometry. In ductile iron castings, silicon is not only a strengthening element. It is also a powerful graphitizing element. It promotes the precipitation of carbon as graphite instead of cementite. A lower silicon content moves the eutectic point and changes the solidification mode. The result is a smaller volume fraction of graphite and a greater tendency for shrinkage porosity.
I also noticed that the silicon content was not being corrected adequately. The original process did not use a separate furnace inoculation step to raise silicon before nodularization. The final silicon was determined mainly by the base iron and the nodulizer. This was inconsistent. To make sound ductile iron castings, I needed a more reliable way to control the final silicon within a narrower range.
Excessive Nodulizer Addition
The second root cause was the nodulizer addition rate. The original process used FeSiMg8RE5 at a rate of 1.6–2.0% by mass of the molten metal. Statistical testing showed that the residual magnesium was often between 0.05% and 0.06%, which was close to the upper control limit. This is too high for this type of ductile iron casting.
High residual magnesium increases shrinkage in ductile iron castings for several reasons. It promotes undercooling and carbide formation at cell boundaries. It increases the number and volume of magnesium-containing inclusions, such as MgO and MgS. These inclusions can obstruct the flow of liquid metal through the narrow channels between dendrites and graphite nodules. In addition, high magnesium tends to reduce the number of effective graphite nuclei if the inoculation is not perfectly controlled. The final solidification becomes more difficult to feed.
The recovery of magnesium can be expressed as:
\[
\eta_{\mathrm{Mg}} = \frac{\mathrm{Mg}_{\mathrm{res}}}{\mathrm{Mg}_{\mathrm{add}}} \times 100
\]
where \(\mathrm{Mg}_{\mathrm{res}}\) is the residual magnesium content and \(\mathrm{Mg}_{\mathrm{add}}\) is the magnesium added with the nodulizer. The measured residual magnesium values showed that magnesium recovery was higher than necessary. I concluded that the nodulizer addition should be reduced.
High Pouring Temperature
The third root cause was the pouring temperature. The original process specified a pouring temperature of 1390–1400 °C. This temperature is relatively high for a wheel hub made of ductile iron. Higher pouring temperature means more superheat. Liquid iron must lose that superheat before solidification can begin. The additional heat input increases liquid contraction and delays solidification in the heavy sections.
The volumetric change caused by superheat can be estimated from:
\[
\Delta V_{\mathrm{superheat}} = \beta V_0 \left(T_{\mathrm{pour}} – T_{\mathrm{liquidus}}\right)
\]
where \(\beta\) is the volume expansion coefficient of liquid iron, \(V_0\) is the original liquid volume, and \(T_{\mathrm{pour}} – T_{\mathrm{liquidus}}\) is the superheat. Reducing the superheat from about 80 °C to about 60 °C reduces the amount of liquid contraction. This helps ductile iron castings solidify with less internal porosity.
Unstable Stream Inoculation
The fourth root cause was related to the pouring machine. The original stream inoculation system used a flow meter that could not accurately control the amount of FeSi75Al1.5 inoculant. In a single mold, some hubs received the correct amount of inoculant, while others received less. This explained why one cavity in the same mold could have shrinkage porosity and another cavity could be sound.
Inoculation is critical for ductile iron castings. It provides heterogeneous nuclei for graphite precipitation. When inoculation is insufficient, the number of graphite nodules decreases, and the graphite morphology can become irregular. The remaining liquid solidifies with a greater degree of constitutional undercooling. This favors the formation of intercellular shrinkage and microporosity.
The target inoculation rate can be written as a mass flow equation:
\[
\dot{m}_{\mathrm{inoc}} = \rho_{\ell} Q_{\ell} C_{\mathrm{inoc}}
\]
where \(\rho_{\ell}\) is the density of the molten iron, \(Q_{\ell}\) is the volumetric metal flow rate, and \(C_{\mathrm{inoc}}\) is the desired inoculant fraction of metal mass. The original flow meter could not maintain a constant \(C_{\mathrm{inoc}}\). As a result, the process was not stable.
Thermal Hot Spots
The fifth root cause was geometric. The wheel hub has thick sections at the top and at the roots of the petal-like projections. These sections have a larger solidification modulus than the surrounding thin walls. The last liquid in the casting is located there. The surrounding metal freezes first and cuts off the feeding path. Because no external chill or local cooling was used, those hot spots remained liquid for a long time and then shrank without being fed.
The solidification modulus is defined as:
\[
M = \frac{V}{A}
\]
where \(V\) is the volume of the section and \(A\) is the cooling surface area. A larger modulus means a longer solidification time. In these ductile iron castings, the modulus of the heavy top section was much larger than the modulus of the adjoining ribs. This created an unfavorable temperature gradient. The hot spot was not fed by gravity because it was the last region to solidify.
Corrective Actions
Based on the root-cause analysis, I introduced five corrective actions. The changes were designed to promote earlier graphite expansion, reduce residual magnesium, improve inoculation stability, lower superheat, and accelerate cooling at the critical hot spots.
Adjusting the Silicon Range
I changed the target silicon content from 2.4–2.6% to 2.6–2.8%. The additional silicon was supplied through an inoculation addition before nodularization or in the treatment ladle, depending on the heat. This served two purposes. First, it increased the final silicon within the new range. Second, it provided an early inoculation effect that increased the number of graphite nuclei.
| Element | Original Range | Modified Range |
|---|---|---|
| C | 3.5–3.7% | 3.5–3.7% |
| Si | 2.4–2.6% | 2.6–2.8% |
| Mn | ≤ 0.5% | ≤ 0.5% |
| S | ≤ 0.03% | ≤ 0.03% |
| P | ≤ 0.07% | ≤ 0.07% |
| Mg residual | 0.03–0.06% | 0.035–0.050% |
| Rare earth | 0.01–0.05% | 0.02–0.04% |
Reducing the Nodulizer Addition
I reduced the FeSiMg8RE5 addition from 1.6–2.0% to 1.4–1.6% by mass of the treated metal. In parallel, I tightened the rare earth range to 0.02–0.04%. The intention was to keep the nodularity above the required level while lowering the shrinkage tendency caused by excessive residual magnesium.
After this change, the residual magnesium settled into a more controllable range of about 0.035–0.050%. I verified the graphite morphology on every trial heat. The graphite remained mostly spheroidal, with nodularity well above the minimum acceptance level. The lower residual magnesium also reduced the number of oxide and sulfide inclusions in the melt. This improved liquid feeding through the final interdendritic channels.
Rebuilding the Stream Inoculation Control System
The next corrective action was to replace the unreliable flow-control equipment on the mechanical pouring machine. I worked with the maintenance team to install a closed-loop dosing system. The new system measures the metal flow rate and adjusts the inoculant feed rate in real time. This kept the inoculant concentration constant throughout the pouring period.
The modified system was verified by performing pour tests and collecting the inoculant delivered over a fixed time. The measured variation was significantly smaller than before. Each cavity in the four-cavity mold now received a consistent amount of FeSi75Al1.5 inoculant. This uniformity was essential for ductile iron castings because the graphite count and shrinkage tendency are strongly affected by late-stage inoculation.
Placing External Chills at Hot Spots
I introduced external chills at the hot spots that had previously produced shrinkage porosity. The chills were made of chilled cast iron and were placed in the mold before closing. Their function was to increase the local cooling rate and reduce the solidification time of the heavy sections.
Using Chvorinov’s rule, the solidification time is:
\[
t_s = B \left( \frac{V}{A} \right)^2
\]
where \(B\) is a mold constant that depends on the mold material and heat-transfer conditions. Chilling the mold locally increases the effective heat extraction, which reduces the value of \(B\) around the hot spot. As a result, the local solidification time becomes closer to the solidification time of the surrounding thinner sections.
| Chill Set | Location | Purpose |
|---|---|---|
| 1# chill | Thick upper boss and heavy pad | Accelerate cooling of the last liquid pool |
| 2# chill | Root transition of the petal-shaped projections | Eliminate the local hot spot at the thick-to-thin transition |
| 3# chill | Side heavy section near the flange | Promote directional solidification toward the feeding path |
The chills changed the temperature field in a favorable way. The thermal gradient \(G\) increased around the former hot spots. This increased the Niyama criterion and reduced the risk of microporosity. I also verified that the chills did not create cold shuts or carbide formation at the surface. The section thickness and pouring temperature were still high enough to avoid those problems.
Controlling the Pouring Temperature
The final corrective action was to control the pouring temperature more strictly. I changed the target from 1390–1400 °C to 1370–1380 °C. This is a moderate reduction, but it decreased the superheat and reduced the liquid contraction. The lower pouring temperature also shortened the time available for heat to be conducted into the mold, which helped the external chills work more effectively.
| Parameter | Before | After |
|---|---|---|
| Nodulizer addition | 1.6–2.0% | 1.4–1.6% |
| Silicon range | 2.4–2.6% | 2.6–2.8% |
| Residual magnesium | 0.05–0.06% | 0.035–0.050% |
| Rare earth | 0.01–0.05% | 0.02–0.04% |
| Stream inoculation | Uncontrolled flow meter | Closed-loop metering system |
| Pouring temperature | 1390–1400 °C | 1370–1380 °C |
| External chills | Not used | Three sets at critical hot spots |
Verification Trials
After implementing all five corrective actions, I prepared a verification batch. The objective was to determine whether the shrinkage porosity had been eliminated in actual production conditions. I used eight consecutive production molds, each containing four hub cavities. This gave a total of 32 ductile iron castings for evaluation.
The castings were allowed to cool completely in the mold. After shake-out, they were cleaned, shot blasted, and visually inspected. I then sectioned every casting through the regions that had previously shown shrinkage porosity. The cut surfaces were polished and examined carefully. None of the 32 castings showed visible shrinkage cavities or internal microporosity in the former defect zones.
| Inspection Item | Result |
|---|---|
| Number of molds | 8 |
| Number of castings produced | 32 |
| Number of castings sectioned | 32 |
| Shrinkage cavities found | 0 |
| Internal microporosity found | 0 |
| Graphite nodularity | Acceptable |
| Carbide formation | Not observed |
The successful verification batch confirmed that the combined changes were effective. I consider this result to be due to the interaction of all five actions rather than any single change. In ductile iron castings, soundness depends on the whole metallurgical and thermal path, not only on the feeding system.
Discussion
The experience gained from these ductile iron castings illustrates an important principle: shrinkage porosity is not always a simple feeding problem. In ductile iron castings, the precipitation of graphite is the main internal source of volume expansion. If that expansion happens early enough and the mold is rigid, it can compensate for the contraction of liquid iron. If graphitization is weak or inoculation is poor, the expansion happens too late or too little. The final liquid remains under pressure and pulls apart, creating voids.
The original process was close to the correct operating window, but it had several small deviations. The silicon content was slightly low. The nodulizer addition was slightly high. The pouring temperature was slightly high. The stream inoculation was unstable. The hot spots were not cooled. Each of these factors alone might not have caused unacceptable scrap. Together, however, they pushed the solidification behavior into the shrinkage-prone regime.
I also learned that external chills are particularly effective for ductile iron castings because they reduce the local solidification time without requiring large risers. By improving the local temperature gradient, the chills made the feeding path remain open longer. The Niyama criterion at the critical section increased enough to avoid microporosity.
The adjustment of silicon is worth discussing in more detail. Silicon is a graphitizing element, but it also increases the quantity of ferrite in the metallic matrix. In the specific hub grade, the elongation and ductility were acceptable. I did not observe any negative effect on tensile strength after the silicon increase. The resulting microstructure was still within specification, with the graphite nodules dispersed uniformly in a ferritic matrix.
Residual magnesium control also deserves attention. Magnesium is necessary for producing spheroidal graphite, but too much magnesium can be harmful. In my experience, the optimum residual magnesium range depends on the section thickness, pouring temperature, and rare earth content. For these ductile iron castings, a residual magnesium range of 0.035–0.050% gave the best balance between nodularity and feeding behavior. The lower nodulizer addition also reduced cost, which is an additional benefit.
Inoculation is the final control point before pouring. A well-designed stream inoculation system ensures that every cavity in the mold has the same number of graphite nuclei. This is particularly important when a mold contains four identical ductile iron castings. If one cavity receives less inoculant than the others, its solidification behavior will be different. The risk of shrinkage porosity will be higher in that cavity. The new closed-loop system eliminated this variability.
Process Recommendations for Future Production
Based on the successful correction, I recommend the following control points for routine production of ductile iron castings in a high-pressure green-sand line:
| Control Point | Recommended Practice |
|---|---|
| Base iron composition | Maintain stable C, Si, Mn, S, P with tight silicon control |
| Nodularization | Use the minimum nodulizer addition that gives required nodularity and residual Mg |
| Inoculation | Use a closed-loop, flow-proportional stream inoculation system |
| Mold chilling | Place external chills at every hot spot identified by section size |
| Pouring temperature | Control the start-of-pour temperature to a narrow range |
| Microstructure verification | Check graphite shape, nodule count, carbide level, and shrinkage porosity |
I also recommend using thermal analysis or online temperature measurement to ensure that the liquid iron entering the pouring station is consistent. The pouring temperature must be measured at the same location in the pouring ladle or pouring cup each time. Small differences in measurement position can create false readings and lead to incorrect decisions.
Conclusion
Shrinkage porosity in ductile iron castings can be solved by a combination of metallurgical and thermal process improvements. In the case of the wheel hub, the main causes were low silicon content, excessive nodulizer addition, high pouring temperature, unstable stream inoculation, and unfavorable hot spots. I addressed each of these causes with a specific corrective action.
The most important changes were increasing the silicon range to 2.6–2.8%, reducing the nodulizer addition to 1.4–1.6%, controlling residual magnesium and rare earth content, rebuilding the inoculation dosing system, installing external chills at critical locations, and lowering the pouring temperature to 1370–1380 °C. The verification run of eight molds produced 32 sectioned hubs without any shrinkage defects.
This work confirmed that ductile iron castings require a holistic process approach. The metallurgy, mold rigidity, thermal gradient, and inoculation practice must be balanced. By addressing all of the relevant variables, it is possible to produce sound ductile iron castings with a high degree of repeatability and low manufacturing cost.
In my own foundry, the same methodology has also been useful for other ductile iron castings with thick sections. The principles of carbon equivalent management, residual magnesium control, stable inoculation, and external chilling are transferable to many ductile iron castings that suffer from shrinkage porosity. I believe that the case described here provides a practical roadmap for any foundry facing similar quality problems.