In our foundry, the production of a commercial vehicle wheel hub in ductile iron casting was repeatedly disturbed by dispersed shrinkage defects. The wheel hub is a safety-critical part, and internal shrinkage porosity is not permitted because it can reduce fatigue strength, pressure tightness, and the reliability of the bearing seats. I initially believed that a well-designed gating and feeding system, supported by solidification simulation, would be sufficient to eliminate the problem. That belief was only partly correct. The gating and feeding system could be optimized, but the real instability came from the molten iron itself, especially from the graphitization behavior after spheroidization and inoculation. By applying thermal analysis to the ductile iron casting process, I was able to identify over-inoculation as the main cause of excessive early graphite expansion and insufficient late graphite expansion. Reducing the in-ladle inoculant addition from 0.4% to 0.3% lowered the dispersed shrinkage rejection rate from more than 12% to less than 2%. This article describes the production problem, the thermal analysis method, the process adjustments, and the results in detail.
1. Wheel Hub Geometry and Technical Requirements
The part I studied is a commercial vehicle front wheel hub produced as a ductile iron casting. It is a rotationally symmetric component with a flange, a central hub barrel, and two bearing connection zones. The flange diameter is approximately 332 mm, the end outer diameter is approximately 152 mm, the total height is approximately 210 mm, and the minimum wall thickness is approximately 7.3 mm. Under the flange, there are ten circular bosses that are later machined into connection holes. The material grade is QT450-10, which requires a combination of tensile strength, elongation, and machinability suitable for a safety-related ductile iron casting.
The two bearing connection zones are the thickest sections of the wheel hub. The first bearing connection zone has a modulus of approximately 1.15 cm, and the second bearing connection zone has a modulus of approximately 0.92 cm. These two zones behave as thermal nodes. Because the wheel hub is a ductile iron casting with multiple dispersed thermal nodes, the feeding distance is limited, and the risk of dispersed shrinkage is high. The main casting difficulty is not a single large hot spot that can be fed by one large feeder. Instead, the difficulty is the presence of several separated hot spots that require both external feeding and self-feeding by graphite expansion.
| Feature | Value |
|---|---|
| Part type | Commercial vehicle front wheel hub |
| Material | QT450-10 ductile iron casting |
| Flange diameter | Approximately 332 mm |
| End outer diameter | Approximately 152 mm |
| Total height | Approximately 210 mm |
| Minimum wall thickness | Approximately 7.3 mm |
| First bearing connection modulus | Approximately 1.15 cm |
| Second bearing connection modulus | Approximately 0.92 cm |
| Critical quality requirement | No internal shrinkage cavity or dispersed shrinkage |
2. Initial Production Process and Feeding Design
The ductile iron casting was produced on a high-pressure green sand molding line with automatic pouring. I used a gating and feeding design based on the balanced solidification principle. In this approach, the external feeder is expected to compensate for liquid and solidification contraction during the early stage, while graphite expansion in the later stage is expected to compensate for the remaining contraction. The success of the method depends strongly on the timing and amount of graphite expansion.
The casting weight was approximately 18.68 kg. The first bearing connection zone had the largest modulus, so I used its modulus as the representative casting modulus:
$$ M_c = 1.15 \text{ cm} $$
The quality perimeter quotient was calculated as:
$$ Q_m = \frac{G}{M_c^3} $$
$$ Q_m = \frac{18.68}{1.15^3} = 12.28 \text{ kg/cm}^3 $$
The shrinkage time fraction was calculated as:
$$ P_c = \frac{1.0}{e^{(0.5M_c + 0.01Q_m)}} $$
$$ P_c = \frac{1.0}{e^{(0.5 \times 1.15 + 0.01 \times 12.28)}} = 0.37 $$
The shrinkage modulus coefficient was calculated as:
$$ f_2 = \sqrt{P_c} $$
$$ f_2 = \sqrt{0.37} = 0.61 $$
The casting shrinkage modulus was calculated as:
$$ M_s = f_2 M_c $$
$$ M_s = 0.61 \times 1.15 = 0.702 \text{ cm} $$
With a feeder balance coefficient \(f_1 = 1.2\) and a feeder pressure coefficient \(f_3 = 1.3\), the required feeder modulus was:
$$ M_R = f_1 f_3 M_s $$
$$ M_R = 1.2 \times 1.3 \times 0.702 = 1.095 \text{ cm} $$
To achieve the required feeding, I placed one side feeder with a 6/9-type exothermic insulating sleeve at the center of each mold cavity. The internal cavity diameter of the sleeve was approximately 60 mm, and its height was approximately 78 mm. With this sleeve, the feeder modulus could reach approximately 1.7 cm, which was expected to be sufficient for the calculated requirement. The sleeve was inserted into the upper mold like a sand core and contacted the lower sand core after mold closing. The sand core had one rectangular opening with dimensions of approximately 30 mm by 22.8 mm. During filling, molten iron flowed through the side opening of the sand core and then through the top opening into the feeder sleeve. The feeder could thus feed the central part of the ductile iron casting through the side opening of the sand core.
For other isolated hot spots, the quality perimeter quotient was greater than 20 kg/cm³. In the early stage of solidification, these hot spots could receive feeding from the feeder. In the later stage, when the feeding channel closed, they were expected to use their own graphite expansion for self-feeding. The gating system was an open system with four cavities per mold. The total sprue area was approximately 13.2 cm², the total runner area was approximately 19 cm², and the total ingate area was approximately 26.8 cm². The area ratio was:
$$ \sum S_{\text{sprue}} : \sum S_{\text{runner}} : \sum S_{\text{ingate}} = 1 : 1.44 : 2 $$
Two ceramic foam filters with dimensions of approximately 75 mm by 75 mm by 22 mm and 10 ppi were placed in the runner. These filters helped stabilize the flow and reduced erosion of the flange cavity. The total weight of the casting and gating and feeding system was approximately 111.8 kg, and the process yield was approximately 62%. The gravity direction was vertically downward. This type of gating and feeding design is common for ductile iron casting wheel hubs produced on clay sand lines. Its advantages include high process yield, a stress concentration notch at the feeder neck, easy feeder removal, and simple grinding of the remaining feeder neck.
| Parameter | Value |
|---|---|
| Casting weight \(G\) | 18.68 kg |
| Representative modulus \(M_c\) | 1.15 cm |
| Quality perimeter quotient \(Q_m\) | 12.28 kg/cm³ |
| Shrinkage time fraction \(P_c\) | 0.37 |
| Shrinkage modulus coefficient \(f_2\) | 0.61 |
| Casting shrinkage modulus \(M_s\) | 0.702 cm |
| Feeder modulus \(M_R\) | 1.095 cm |
| Feeder sleeve type | 6/9 exothermic insulating sleeve |
| Feeder sleeve internal diameter | 60 mm |
| Feeder sleeve height | 78 mm |
| Feeder modulus achieved | Approximately 1.7 cm |
| Gating area ratio | Sprue : runner : ingate = 1 : 1.44 : 2 |
| Process yield | Approximately 62% |
3. Simulation and Initial Trial Results
I used casting simulation software to validate the initial gating and feeding process. The simulation predicted isolated liquid phase regions and shrinkage porosity in the flange machining holes, the first bearing connection zone, and the second bearing connection zone. In the simulation, the inoculation effect was set to “good” and the graphite precipitation parameter was set to 8. These settings represented a normal inoculation condition for ductile iron casting. The predicted shrinkage defects were consistent with the production risk, but the simulation also indicated that a stronger inoculation effect and more graphite precipitation could improve self-feeding.
Based on this result, I modified the process. The ingate positions were moved so that each of the four ingates was aimed at the circular bosses under the flange. In the updated simulation, the inoculation effect was set to “very good” and the graphite precipitation parameter was set to 10. Under these assumptions, the simulation predicted no shrinkage cavity or dispersed shrinkage in the ductile iron casting. This encouraged me to control the carbon equivalent and to strengthen inoculation so that the iron liquid would have a higher self-feeding capacity. The goal was to produce a wheel hub ductile iron casting free of internal shrinkage defects.
I then manufactured the improved lower pattern plate and upper pattern plate and began trials. The molten iron was melted in a medium-frequency induction furnace. The charge composition was controlled as follows: 3.65% to 3.75% C, 1.95% to 2.05% Si, 0.4% to 0.5% Mn, less than 0.05% P, and less than 0.025% S. About 1.2 t of molten iron was transferred to a wire-feeding spheroidization ladle. A cored wire containing approximately 30% Mg was used for spheroidization, and the wire feed length was approximately 21.5 m. After spheroidization, the molten iron was transferred to a pouring ladle for inoculation. The bottom of the pouring ladle was filled with approximately 0.4% of a silicon-calcium-barium inoculant containing 4% to 6% Ba and with a particle size of 3 to 8 mm. During pouring into the mold cavity, an additional stream inoculant of silicon-calcium-barium with a particle size of 0.2 to 0.7 mm was added at approximately 0.1%.
After spheroidization and inoculation, the chemical composition of the molten iron was controlled as follows: 3.50% to 3.60% C, 2.50% to 2.60% Si, 0.4% to 0.5% Mn, less than 0.05% P, less than 0.02% S, and 0.032% to 0.042% Mg. The final carbon equivalent was controlled in the range of 4.33% to 4.47%. The time from the start of spheroidization to the end of pouring was controlled within 9 minutes. The pouring temperature was 1,350 to 1,400 °C.
Using the improved gating and feeding system, the melting charge, and the spheroidization and inoculation process, I produced 9 molds with a total of 36 wheel hub castings. I sectioned the first mold and the last mold, a total of 8 castings, and found no shrinkage cavity or dispersed shrinkage. All castings were drilled at the centers of the ten bosses under the flange. The flange surface, the first bearing connection zone, and the second bearing connection zone were machined to a depth of approximately 3 mm. No shrinkage cavity or dispersed shrinkage was found on the machined surfaces. I then melted one furnace of molten iron in a 6 t medium-frequency induction furnace for a small batch production. Five pouring ladles were tapped, each about 1,200 kg. A total of 35 molds were poured, producing 140 wheel hub castings. After machining, one casting was found with an internal shrinkage cavity. The dispersed shrinkage rejection rate was approximately 0.714%. However, during large-scale production, many wheel hub castings were found with dispersed shrinkage at the first bearing connection zone after machining. The rejection rate due to dispersed shrinkage after machining reached more than 12%. The defect appeared as a ring-shaped dispersed shrinkage on the section of the first bearing connection zone.
| Production Stage | Molds | Castings | Defect Observation | Rejection Rate |
|---|---|---|---|---|
| Initial trial | 9 | 36 | No shrinkage in sectioned and machined samples | Not applicable |
| Small batch | 35 | 140 | One casting with internal shrinkage cavity | Approximately 0.714% |
| Large-scale production | Many | Many | Dispersed shrinkage at first bearing connection | More than 12% |
4. Thermal Analysis of the Molten Iron
The large difference between the small batch result and the large-scale production result showed that the process was not stable. The simulation could not accurately predict the effect of graphite expansion under different melting, spheroidization, and inoculation conditions. The time and magnitude of graphite expansion in ductile iron casting are difficult to calculate. Simulation software can only approximate graphite expansion using methods such as density curve analysis. Therefore, the simulation result and the actual production result can differ significantly.
To study the graphite precipitation and expansion behavior during solidification, I used a thermal analyzer to record the solidification characteristic curve of the molten iron. The thermal analysis equipment divides solidification into four stages: the primary stage, which may include primary austenite or primary graphite; the early eutectic stage; the stable eutectic graphite precipitation stage, also called secondary graphite precipitation; and the final solidification stage, which is characterized by a thermal conductivity coefficient. In the production line for large-scale wheel hub ductile iron casting, when the molten iron had just been poured into the pouring ladle, I took a sample from the top of the ladle and poured it into a rapid thermal analysis sample cup. The sample cup already contained 0.1% of a silicon-calcium-barium inoculant with a particle size of 0.2 to 0.7 mm. The purpose of this small-particle inoculant in the cup was to simulate the stream inoculation effect on the production line so that the iron state in the cup would be as close as possible to the iron state in the mold cavity.
The thermal analysis result for the mass production iron showed that the iron was eutectic solidification without primary austenite. This indicated that the casting was not likely to form macro shrinkage cavities and that the carbon and silicon composition control was reasonable. However, the recalescence temperature reached 9.2 °C. The ideal recalescence for this type of ductile iron casting is 2 to 7 °C. A recalescence of 9.2 °C indicated that too much graphite was precipitated in the early eutectic stage. The number of graphite nuclei was excessive, which consumed too much sulfur and oxygen from the molten iron. As a result, the sulfur and oxygen concentration in the later solidification stage was too low, and the number of active nuclei decreased. The thermal conductivity coefficient, which characterizes graphite precipitation at the final solidification stage, was 20. The ideal value is 22 to 28. A value of 20 indicated that the amount of graphite precipitated in the final solidification zone was low, and the self-feeding ability of the isolated liquid regions was weak.
Based on my experience, both insufficient inoculation of hypoeutectic iron and over-inoculation of eutectic iron can cause a high recalescence temperature. The thermal analysis result for the large-scale production iron showed that the iron was eutectic and that over-inoculation was likely. Therefore, I adjusted the next ladle after wire-feeding spheroidization. The amount of silicon-calcium-barium inoculant added to the pouring ladle was reduced from 0.4% to 0.3%. After spheroidization, the molten iron was poured into the pouring ladle, and a top sample was taken and poured into a rapid thermal analysis cup. The cup already contained 0.1% of the same small-particle silicon-calcium-barium inoculant. The new thermal analysis result showed that the recalescence temperature decreased to 6.5 °C and the thermal conductivity coefficient increased to 24. The recalescence decreased by 2.7 °C, and the thermal conductivity coefficient increased by 4 units. This indicated that early graphite nucleation and precipitation decreased, while later graphite nucleation increased. This change was favorable for the self-feeding ability of the ductile iron casting and was expected to significantly reduce the probability of dispersed shrinkage.
| Thermal Analysis Parameter | Before Adjustment | After Adjustment | Ideal Range |
|---|---|---|---|
| Solidification type | Eutectic | Eutectic | Eutectic for this part |
| Primary austenite | None | None | None |
| Recalescence temperature | 9.2 °C | 6.5 °C | 2 to 7 °C |
| Thermal conductivity coefficient | 20 | 24 | 22 to 28 |
| In-ladle inoculant addition | 0.4% | 0.3% | Optimized by thermal analysis |
| Stream inoculant addition | 0.1% | 0.1% | 0.1% |
5. Root Cause Analysis
The root cause of the dispersed shrinkage in the ductile iron casting was over-inoculation. Over-inoculation caused excessive graphite precipitation in the early stage of solidification. The early graphite expansion was too large. This early expansion consumed the available carbon and reduced the amount of carbon remaining for later graphite precipitation. It also consumed sulfur and oxygen, which are necessary for nucleation. As a result, the final solidification zone did not have enough active nuclei for late graphite expansion. The late graphite expansion was therefore insufficient to compensate for the solidification contraction of the isolated liquid regions. The first bearing connection zone, which had the largest modulus and the longest solidification time, was the most sensitive region.
The volumetric behavior of the ductile iron casting can be described qualitatively by the following relationship. The total volume change during solidification is the sum of liquid contraction, solidification contraction, and graphite expansion:
$$ \Delta V_{\text{total}} = \Delta V_{\text{liquid contraction}} + \Delta V_{\text{solidification contraction}} + \Delta V_{\text{graphite expansion}} $$
For self-feeding to be effective, the graphite expansion must occur at a time and location where it can compensate for the contraction of the isolated liquid region. If graphite expansion occurs too early, it may only pressurize the already rigid shell and may not feed the later-solidifying hot spot. If graphite expansion is distributed too early and is not sustained, the later stage lacks the expansion needed for self-feeding. The thermal analysis parameters reflect this behavior:
$$ \Delta T_r = T_{\text{max}} – T_{\text{min}} $$
where \(\Delta T_r\) is the recalescence temperature, \(T_{\text{max}}\) is the maximum eutectic temperature, and \(T_{\text{min}}\) is the minimum eutectic temperature before recalescence. A high \(\Delta T_r\) indicates strong early graphite precipitation. In addition, the thermal conductivity coefficient can be used as an indicator of late graphite precipitation:
$$ K_{\text{th}} = f(t_{\text{final}}, \text{graphite precipitation}) $$
A low \(K_{\text{th}}\) indicates weak late graphite precipitation. In our case, the original iron had \(\Delta T_r = 9.2 °C\) and \(K_{\text{th}} = 20\). After reducing the inoculant, the iron had \(\Delta T_r = 6.5 °C\) and \(K_{\text{th}} = 24\). The change in these two parameters was directly linked to the reduction of dispersed shrinkage in the ductile iron casting.
The carbon equivalent also affects graphite precipitation. For ductile iron casting, the carbon equivalent can be expressed as:
$$ CE = C + \frac{Si + P}{3} $$
In our production, the final carbon equivalent was controlled between 4.33% and 4.47%. This range was selected to promote graphitization and self-feeding. However, carbon equivalent alone was not sufficient. The inoculation process determined how the carbon was precipitated. Over-inoculation caused the carbon to precipitate too early, while under-inoculation could cause chill and insufficient graphite expansion. The thermal analysis method provided a way to balance these effects.
| Cause | Effect on Solidification | Consequence for Ductile Iron Casting |
|---|---|---|
| Over-inoculation | Excessive early graphite nucleation | High recalescence, early expansion, late graphite starvation |
| Excess early graphite precipitation | Consumption of C, S, and O | Fewer nuclei in final solidification zone |
| Insufficient late graphite expansion | Weak self-feeding pressure | Dispersed shrinkage at first bearing connection |
| Large modulus at first bearing connection | Long solidification time | Most sensitive to feeding shortage |
| Simulation uncertainty | Graphite expansion not accurately modeled | Simulation may predict no defects while production shows defects |
6. Corrective Actions and Process Optimization
The corrective action was to reduce the in-ladle inoculant addition from 0.4% to 0.3% while keeping the stream inoculant at 0.1%. The purpose was to reduce early graphite nucleation and precipitation, lower the recalescence temperature, and preserve more nucleation sites for the final solidification stage. The spheroidization treatment, magnesium content, pouring temperature, and pouring time were kept within the original control ranges. The gating and feeding system was not changed because the thermal analysis clearly showed that the problem was related to the iron treatment rather than the mold filling and feeding design.
After the adjustment, the molten iron was tested again with the thermal analyzer. The recalescence temperature decreased from 9.2 °C to 6.5 °C, and the thermal conductivity coefficient increased from 20 to 24. These values were closer to the ideal ranges. The iron produced with the adjusted inoculation process was used to cast the wheel hub ductile iron casting. After machining, no shrinkage cavity or dispersed shrinkage was found in the first trial. The process was then applied to multiple large-scale production batches.
The following formula can be used to evaluate the improvement in self-feeding capability:
$$ \eta_{\text{self}} = \frac{\Delta V_{\text{late graphite}}}{\Delta V_{\text{contraction, late}}} $$
where \(\eta_{\text{self}}\) is the self-feeding efficiency, \(\Delta V_{\text{late graphite}}\) is the late graphite expansion volume, and \(\Delta V_{\text{contraction, late}}\) is the late solidification contraction volume. When \(\eta_{\text{self}} \geq 1\), the late graphite expansion can fully compensate for the late contraction. In practice, the thermal conductivity coefficient and recalescence temperature are used as indirect indicators of \(\eta_{\text{self}}\). Reducing the inoculant increased \(K_{\text{th}}\) and decreased \(\Delta T_r\), which increased \(\eta_{\text{self}}\) and reduced dispersed shrinkage.
| Process Variable | Original Condition | Optimized Condition | Reason for Change |
|---|---|---|---|
| In-ladle inoculant | 0.4% Si-Ca-Ba | 0.3% Si-Ca-Ba | Reduce early graphite precipitation |
| Stream inoculant | 0.1% Si-Ca-Ba | 0.1% Si-Ca-Ba | Maintain final nucleation |
| Recalescence | 9.2 °C | 6.5 °C | Bring into ideal range |
| Thermal conductivity coefficient | 20 | 24 | Improve late graphite precipitation |
| Gating and feeding | Balanced solidification design | Unchanged | Problem was in iron treatment |
| Pouring temperature | 1,350 to 1,400 °C | 1,350 to 1,400 °C | Maintain stable filling |
| Pouring time after spheroidization | Within 9 minutes | Within 9 minutes | Prevent magnesium fade and inoculation fade |
7. Production Results After Optimization
After reducing the in-ladle inoculant addition, the dispersed shrinkage rejection rate of the wheel hub ductile iron casting decreased significantly. In one production batch, 2,716 castings were poured. In addition, 23 castings from a previous missed inspection were included, so a total of 2,739 castings were inspected. The number of castings rejected due to dispersed shrinkage was 24, giving a dispersed shrinkage rejection rate of approximately 0.88%. The total rejection rate from all defects, including sand holes, dispersed shrinkage, gas holes, and handling damage, was approximately 3.91%. The company requirement is that the rejection rate due to shrinkage cavity and dispersed shrinkage in ductile iron casting must be controlled within 2%, and the total rejection rate after machining must be controlled within 5%. The optimized process met these requirements.
| Batch Item | Value |
|---|---|
| Product type | Commercial vehicle wheel hub ductile iron casting |
| Single casting weight | 18.68 kg |
| Poured quantity | 2,716 castings |
| Poured tonnage | 50.73 t |
| Inspected quantity | 2,739 castings |
| Inspected tonnage | 51.16 t |
| Warehouse quantity | 2,714 castings |
| Warehouse tonnage | 50.7 t |
| Dispersed shrinkage rejects | 24 castings |
| Dispersed shrinkage rejection rate | Approximately 0.88% |
| Total rejects | 107 castings |
| Total rejection rate | Approximately 3.91% |
| Requirement for shrinkage and dispersed shrinkage | Less than 2% |
| Requirement for total rejection after machining | Less than 5% |
The production results confirmed that the thermal analysis method was effective for controlling the ductile iron casting process. The reduction in in-ladle inoculant did not cause chill, graphite degeneration, or other defects. The stream inoculant at 0.1% was sufficient to provide final nucleation. The combination of the balanced solidification gating and feeding design and the optimized inoculation process produced a stable ductile iron casting process with a low dispersed shrinkage rejection rate.
8. Discussion
The experience with this wheel hub ductile iron casting shows that gating and feeding system optimization alone cannot guarantee freedom from dispersed shrinkage. The graphite expansion behavior of ductile iron casting is sensitive to melting charge, spheroidization, inoculation, and cooling rate. When the inoculation is excessive, the early graphite expansion may appear beneficial in simulation because it increases the total expansion. However, in actual production, the early expansion may not feed the later-solidifying hot spots. It may also exhaust the nucleation potential needed for late expansion. Therefore, the timing of graphite expansion is as important as the total amount.
The thermal analysis method allows the foundry to observe the solidification curve and to quantify the recalescence temperature and thermal conductivity coefficient. These parameters can be used to judge whether the inoculation level is appropriate. For the ductile iron casting in this study, the ideal recalescence was 2 to 7 °C, and the ideal thermal conductivity coefficient was 22 to 28. The original iron had a recalescence of 9.2 °C and a thermal conductivity coefficient of 20, which indicated over-inoculation and weak late graphite precipitation. After reducing the in-ladle inoculant, the recalescence decreased to 6.5 °C and the thermal conductivity coefficient increased to 24, which indicated a more balanced graphite precipitation process.
The balanced solidification principle assumes that the feeder compensates for early contraction and that graphite expansion compensates for late contraction. This assumption is valid only when graphite expansion occurs at the right time and in the right location. The thermal analysis parameters provide a practical way to monitor this condition. If the recalescence is too high, the early graphite expansion is too strong, and the late expansion may be insufficient. If the recalescence is too low, the iron may be under-inoculated, and the risk of chill and insufficient graphite expansion increases. The optimal window must be determined for each ductile iron casting and each production line.
In addition to the thermal analysis parameters, the carbon equivalent and the magnesium content must be controlled. In this production, the carbon equivalent was 4.33% to 4.47%, and the magnesium content after spheroidization was 0.032% to 0.042%. These ranges were suitable for the wheel hub ductile iron casting. The pouring time after spheroidization was kept within 9 minutes to avoid magnesium fade and inoculation fade. The pouring temperature was 1,350 to 1,400 °C to ensure smooth filling without excessive sand burn-on or cold shut.
The production batch statistics showed that the optimized process achieved a dispersed shrinkage rejection rate of less than 2% and a total rejection rate of less than 5%. The improvement was achieved without changing the gating and feeding system, which confirms that the root cause was over-inoculation. The thermal analysis method was faster and more direct than trial-and-error adjustments. It also reduced the risk of producing large quantities of defective ductile iron castings before the problem was detected.
| Parameter or Practice | Original Approach | Optimized Approach | Effect on Ductile Iron Casting |
|---|---|---|---|
| Feeding design | Balanced solidification with side feeder | Unchanged | Provides early feeding, relies on late self-feeding |
| Inoculation control | Fixed 0.4% in-ladle | 0.3% in-ladle plus 0.1% stream | Balances early and late graphite precipitation |
| Thermal analysis | Not routinely used | Used for each critical ladle | Detects over-inoculation and late graphite deficiency |
| Simulation | Used with assumed inoculation and graphite parameters | Used as supporting tool, not sole control | Reduces uncertainty from graphite expansion modeling |
| Scrap rate | More than 12% dispersed shrinkage | Approximately 0.88% dispersed shrinkage | Significant improvement in ductile iron casting quality |
9. Practical Guidelines for Thermal Analysis Control
Based on this work, I recommend the following practical guidelines for using thermal analysis to reduce dispersed shrinkage in ductile iron casting:
- Take the thermal analysis sample from the pouring ladle after spheroidization and in-ladle inoculation. The sample should represent the iron that will enter the mold cavity.
- Add the same type and amount of stream inoculant to the sample cup as used in production. This makes the sample cup condition closer to the mold filling condition.
- Record the recalescence temperature and thermal conductivity coefficient for every critical ladle. Compare the values with the ideal ranges for the specific ductile iron casting.
- If the recalescence is above the ideal range and the thermal conductivity coefficient is below the ideal range, reduce the in-ladle inoculant addition. This indicates over-inoculation and insufficient late graphite precipitation.
- If the recalescence is below the ideal range and the thermal conductivity coefficient is below the ideal range, do not reduce the inoculant. This may indicate under-inoculation or a problem with the spheroidization treatment.
- Keep the stream inoculant at a sufficient level to provide final nucleation. In this work, 0.1% stream inoculant was effective when the in-ladle inoculant was reduced to 0.3%.
- Monitor the carbon equivalent, magnesium content, pouring temperature, and pouring time. Thermal analysis cannot replace good melting and pouring practice.
- Use thermal analysis together with simulation and sectioning. Simulation is useful for gating and feeding design, but it is not a substitute for actual graphite expansion control.
The thermal analysis method is not limited to the wheel hub ductile iron casting. It can be applied to other ductile iron casting parts with dispersed thermal nodes, such as differential carriers, brake calipers, steering knuckles, and gearbox housings. The key is to establish the relationship between the thermal analysis parameters and the defect sensitivity of the specific ductile iron casting.
10. Conclusions
1. The use of balanced solidification theory and casting simulation for the gating and feeding design of ductile iron casting is meaningful. However, the melting charge, spheroidization process, and inoculation process strongly affect graphite precipitation during solidification. Controlling the graphite precipitation process so that it is stable and long-lasting is the key to self-feeding and to making simulation results consistent with actual production results.
2. Thermal analysis equipment can accurately record the temperature curve of molten iron during solidification. From this curve, the key heat-release stages and corresponding parameters can be obtained. The ductile iron casting iron in this study was eutectic, without primary austenite or primary graphite, so the curve showed only three stages. By reducing the graphite heat release in the early eutectic stage, that is, by reducing the recalescence, the early graphite precipitation and volume expansion were reduced. This preserved more graphite nucleation sites for the later stages. The sustained graphite precipitation improved the self-feeding ability of the iron and reduced the dispersed shrinkage rejection rate of the wheel hub ductile iron casting.
3. Adding an excessive amount of inoculant not only wastes material and increases the production cost of ductile iron casting, but can also cause internal dispersed shrinkage. Using thermal analysis to compare the solidification curves of ductile iron casting iron with different inoculant additions allows the foundry to quickly judge whether the inoculant addition is reasonable.
4. In this production case, reducing the in-ladle inoculant addition from 0.4% to 0.3% reduced the recalescence from 9.2 °C to 6.5 °C and increased the thermal conductivity coefficient from 20 to 24. The dispersed shrinkage rejection rate decreased from more than 12% to approximately 0.88% in the reported batch, and the total rejection rate was approximately 3.91%. The optimized process met the company requirements for ductile iron casting quality.
5. The thermal analysis method is a practical process control tool for ductile iron casting. It helps the foundry move from reactive scrap sorting to proactive iron treatment control. When it is combined with a robust gating and feeding design, consistent melting practice, and controlled pouring, it can significantly improve the stability of ductile iron casting production.
| Final Summary | Result |
|---|---|
| Main defect in ductile iron casting | Dispersed shrinkage at first bearing connection of wheel hub |
| Root cause | Over-inoculation causing excessive early graphite expansion and insufficient late graphite expansion |
| Detection method | Thermal analysis of molten iron before pouring |
| Key parameters | Recalescence temperature and thermal conductivity coefficient |
| Corrective action | Reduce in-ladle inoculant from 0.4% to 0.3% |
| Before optimization | More than 12% dispersed shrinkage rejection |
| After optimization | Approximately 0.88% dispersed shrinkage rejection in the reported batch |
| Total rejection after optimization | Approximately 3.91% |
| Overall conclusion | Thermal analysis effectively reduced scrap in ductile iron casting |