I have worked for many years with ductile iron castings, especially safety-critical wheel hub components for commercial vehicles. In my experience, shrinkage porosity and dispersed shrinkage are among the most persistent defects in ductile iron castings because the volume changes during solidification are controlled by both liquid contraction and graphite expansion. When the graphite expansion occurs too early, the feeding channels can close before the last liquid regions have been fed. When graphite expansion occurs too late or is too weak, the last solidifying regions cannot self-feed. This balance is difficult to predict with gating and feeding design alone. Therefore, I turned to thermal analysis of the molten iron before pouring. The purpose was not merely to check chemistry, but to observe the solidification path of the ductile iron castings in real time and to identify whether the inoculation practice was causing excessive early graphite precipitation and insufficient late graphite expansion.
The component I studied was a commercial vehicle wheel hub casting, a rotationally symmetric part with a flange, a central hub, and two bearing connection zones. The material was QT450-10 ductile iron, which requires both high strength and sufficient elongation. Because the wheel hub carries vehicle load and is a safety-related part, internal shrinkage cavities and dispersed shrinkage are not acceptable. The main challenge was that the casting had several scattered thermal nodes, and the feeding path was not simple. I found that improving the gating and feeding system alone could reduce some defects but could not eliminate the internal shrinkage in ductile iron castings produced on a high-volume molding line. The rejection rate caused by dispersed shrinkage remained above 12% in mass production. This was the problem that led me to apply thermal analysis to the molten iron and to optimize the inoculation process.
1. Casting Geometry and Technical Requirements
The wheel hub casting weighed approximately 18.68 kg. Its flange diameter was about 332 mm, the outer diameter of the end head was about 152 mm, and the total height was about 210 mm. The minimum wall thickness was about 7.3 mm. Below the flange, which connects to the brake system, there were ten circular bosses. These bosses were later drilled to form connection holes. The first and second bearing connection zones were thicker, approximately 24 mm. The thermal modulus of the first bearing connection zone was about 1.15 cm, and the thermal modulus of the second bearing connection zone was about 0.92 cm. Because these zones were thick and isolated, they were the most likely locations for shrinkage defects in the ductile iron castings.
| Parameter | Value | Unit |
|---|---|---|
| Casting type | Commercial vehicle wheel hub | — |
| Material | QT450-10 ductile iron | — |
| Casting weight | 18.68 | kg |
| Flange diameter | 332 | mm |
| End head outer diameter | 152 | mm |
| Total height | 210 | mm |
| Minimum wall thickness | 7.3 | mm |
| First bearing connection modulus | 1.15 | cm |
| Second bearing connection modulus | 0.92 | cm |
| Number of flange bosses | 10 | — |
| Required internal quality | No shrinkage cavity or dispersed shrinkage | — |
The production line was a high-pressure green sand molding line with automatic pouring. The molding process used horizontal flaskless molding, and the casting was produced in a four-cavity layout. The initial gating and feeding design used balanced solidification technology. I calculated the modulus of the casting, the quality perimeter quotient, the shrinkage time fraction, the shrinkage modulus coefficient, the casting shrinkage modulus, and the feeder modulus. These calculations are summarized below.
The quality perimeter quotient was calculated from the casting weight and the controlling modulus:
$$ Q_m = \frac{G}{M_c^3} = \frac{18.68}{1.15^3} = 12.28 \ \text{kg/cm}^3 $$
where \( G \) is the casting weight in kg and \( M_c \) is the controlling modulus in cm. The shrinkage time fraction was calculated as:
$$ P_c = \frac{1.0}{e^{(0.5M_c + 0.01Q_m)}} = 0.37 $$
The shrinkage modulus coefficient was taken as the square root of the shrinkage time fraction:
$$ f_2 = \sqrt{P_c} = \sqrt{0.37} = 0.61 $$
The casting shrinkage modulus was then:
$$ M_s = f_2 M_c = 0.61 \times 1.15 = 0.702 \ \text{cm} $$
The feeder modulus was calculated using the feeder balance coefficient \( f_1 = 1.2 \) and the feeder pressure coefficient \( f_3 = 1.3 \):
$$ M_R = f_1 f_3 M_s = 1.2 \times 1.3 \times 0.702 = 1.095 \ \text{cm} $$
Based on these calculations, I selected an exothermic insulating feeder sleeve with a nominal cavity diameter of 60 mm and a height of 78 mm. The feeder sleeve was designed to achieve a feeder modulus of approximately 1.7 cm. It was placed in the center of each casting and connected to the casting through a sand core. The sand core had a rectangular opening of 30 mm by 22.8 mm, allowing liquid iron to enter the feeder and allowing the feeder to feed the central region of the casting. The gating system was an open system with a pouring cup, a sprue, a runner, and ingates. The total sprue area was about 13.2 cm², the total runner area was about 19 cm², and the total ingate area was about 26.8 cm². The ratio of sprue to runner to ingate was approximately 1:1.44:2. Two ceramic foam filters of 75 mm × 75 mm × 22 mm with 10 ppi were placed in the runner to ensure smooth filling and to reduce erosion of the flange cavity. The total weight of the casting and gating system was about 111.8 kg, and the process yield was about 62%.
| Design Parameter | Symbol | Value | Unit |
|---|---|---|---|
| Casting weight | G | 18.68 | kg |
| Controlling 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 balance coefficient | f_1 | 1.2 | — |
| Feeder pressure coefficient | f_3 | 1.3 | — |
| Feeder modulus required | M_R | 1.095 | cm |
| Feeder modulus achieved | M_R,actual | 1.7 | cm |
| Sprue area | — | 13.2 | cm² |
| Runner area | — | 19 | cm² |
| Ingate area | — | 26.8 | cm² |
| Gating ratio | — | 1:1.44:2 | — |
| Process yield | — | 62 | % |
2. Initial Simulation and Production Problem
I used casting simulation software to verify the initial gating and feeding design. The simulation predicted isolated liquid regions in the flange area, the first bearing connection zone, and the second bearing connection zone. These isolated regions were potential sites for shrinkage cavities and dispersed shrinkage in the ductile iron castings. In the simulation, I set the inoculation effect to a moderate level and the graphite precipitation parameter to a moderate value. Under these conditions, the simulation showed a risk of defects.
Later, I modified the ingate positions so that each of the four ingates was aimed at the circular bosses below the flange. I also changed the simulation settings to a very good inoculation effect and a higher graphite precipitation parameter. With these changes, the simulation predicted no shrinkage defects. This result suggested that if the molten iron had excellent inoculation and strong graphite expansion, the casting could self-feed and avoid shrinkage. Based on this, I decided to control the carbon equivalent and to strengthen the inoculation treatment to improve the self-feeding ability of the ductile iron castings.
I produced trial castings using the modified gating system and an improved melting and inoculation practice. The base iron was melted in a medium-frequency induction furnace. The charge composition was controlled to 3.65%–3.75% C, 1.95%–2.05% Si, 0.4%–0.5% Mn, less than 0.05% P, and less than 0.025% S. The molten iron was transferred to a treatment ladle and nodularized using a cored wire containing 30% Mg. The wire length was about 21.5 m. After nodularization, the iron was transferred to a pouring ladle for inoculation. The pouring ladle had a bottom layer of 0.4% silicon-calcium-barium inoculant with 4%–6% Ba and a particle size of 3–8 mm. During pouring, a stream inoculant of 0.1% silicon-calcium-barium with a particle size of 0.2–0.7 mm was added. The final chemistry after nodularization and inoculation was 3.50%–3.60% C, 2.50%–2.60% Si, 0.4%–0.5% Mn, less than 0.05% P, less than 0.02% S, and 0.032%–0.042% Mg. The carbon equivalent was controlled in the range of 4.33%–4.47%. The time from the start of nodularization to the end of pouring was kept within 9 minutes. The pouring temperature was 1,350–1,400 °C.
| Stage | C (%) | Si (%) | Mn (%) | P (%) | S (%) | Mg (%) |
|---|---|---|---|---|---|---|
| Base iron charge | 3.65–3.75 | 1.95–2.05 | 0.4–0.5 | <0.05 | <0.025 | — |
| After nodularization and inoculation | 3.50–3.60 | 2.50–2.60 | 0.4–0.5 | <0.05 | <0.02 | 0.032–0.042 |
| Carbon equivalent | 4.33–4.47% | |||||
| Time from nodularization to pouring | ≤9 min | |||||
| Pouring temperature | 1,350–1,400 °C | |||||
In a small trial, I produced nine molds with four castings each, for a total of 36 wheel hub castings. I sectioned the first and last molds and inspected eight castings. I did not find shrinkage cavities or dispersed shrinkage. I also drilled the ten bosses on the flange face and machined the flange face, the first bearing connection zone, and the second bearing connection zone to a depth of about 3 mm. The machined surfaces showed no shrinkage defects. Then I melted one furnace of iron in a 6-ton medium-frequency induction furnace and produced five pouring ladles, each about 1,200 kg. I poured 35 molds and produced 140 wheel hub castings. After machining, I found one casting with an internal shrinkage cavity. The dispersed shrinkage rejection rate was about 0.714%. This result was encouraging.
However, when the process was transferred to large-scale production, the rejection rate caused by dispersed shrinkage in the first bearing connection zone increased sharply. After machining, many wheel hub castings showed a ring-shaped dispersed shrinkage defect on the section of the first bearing connection zone. The rejection rate exceeded 12%. This was unacceptable for ductile iron castings used in safety-critical applications. The contradiction between the small trial and mass production indicated that the process was not stable. The gating and feeding design was fixed, but the molten iron condition was changing. I needed a method to measure the solidification characteristics of the molten iron in real time and to identify the root cause.
3. Thermal Analysis of the Molten Iron
I used a commercial cast iron thermal analysis instrument to record the cooling curve and the solidification parameters of the molten iron. The instrument divided the solidification process into four stages: the initial stage, the early eutectic stage, the stable eutectic graphite precipitation stage, and the final solidification stage. The initial stage corresponds to the formation of primary austenite or primary graphite. The early eutectic stage corresponds to the first burst of eutectic graphite precipitation. The stable eutectic stage corresponds to the steady precipitation of secondary graphite. The final solidification stage corresponds to the last liquid regions, where the thermal conductivity coefficient is used as an indicator of graphite precipitation at the end of solidification.
| Solidification Stage | Physical Meaning | Key Thermal Parameter | Practical Significance for Ductile Iron Castings |
|---|---|---|---|
| Initial stage | Formation of primary austenite or primary graphite | Liquidus temperature, primary undercooling | Indicates whether the iron is hypoeutectic, eutectic, or hypereutectic |
| Early eutectic stage | First rapid eutectic graphite precipitation | Recalescence temperature, maximum eutectic temperature | High recalescence means excessive early graphite expansion |
| Stable eutectic stage | Steady precipitation of secondary graphite | Eutectic plateau, cooling rate | Controls the duration of expansion and self-feeding |
| Final solidification stage | Last liquid regions solidify | Thermal conductivity coefficient | Low value means insufficient late graphite expansion |
For the thermal analysis sample, I took molten iron from the top of the pouring ladle after nodularization and inoculation. I poured it into a quick thermal analysis cup. The cup already contained 0.1% silicon-calcium-barium inoculant with a particle size of 0.2–0.7 mm. The purpose of this small-particle inoculant was to simulate the stream inoculation effect on the production line, so that the sample cup condition was as close as possible to the condition of the molten iron entering the mold cavity. This is important because the thermal analysis result must represent the actual solidification behavior of the ductile iron castings in the mold.
The first thermal analysis result from mass production is summarized below. The molten iron solidified as a eutectic alloy without primary austenite. This indicated that the carbon and silicon contents were controlled reasonably and that the castings were not likely to form large macro-shrinkage cavities. However, the recalescence temperature reached 9.2 °C. The ideal recalescence range for ductile iron castings in my experience is 2–7 °C. A recalescence of 9.2 °C indicated that the amount of graphite precipitated in the early eutectic stage was too high. In other words, the early graphite expansion was excessive. The thermal conductivity coefficient was 20, while the ideal range was 22–28. A thermal conductivity coefficient of 20 indicated that graphite precipitation in the final solidification stage was insufficient. Therefore, the last liquid regions had a weak self-feeding ability.
| Thermal Parameter | Measured Value | Ideal Range | Interpretation |
|---|---|---|---|
| Solidification type | Eutectic | Eutectic or slightly hypoeutectic | No primary austenite, low macro-shrinkage risk |
| Recalescence temperature | 9.2 °C | 2–7 °C | Too high; excessive early graphite precipitation |
| Thermal conductivity coefficient | 20 | 22–28 | Too low; insufficient late graphite expansion |
| Expected effect on defects | High dispersed shrinkage risk | Low dispersed shrinkage risk | Early expansion closes feeding channels; late expansion is insufficient |
The mechanism can be explained by the sequence of graphite precipitation and volume change. When the molten iron is over-inoculated, many graphite nuclei form at the beginning of eutectic solidification. These nuclei grow rapidly and release a large amount of latent heat, causing a high recalescence. The rapid graphite growth also causes early volume expansion. This early expansion can push the remaining liquid away from the feeding channel and can close the feeding path between the feeder and the isolated liquid regions. At the same time, the rapid consumption of sulfur and oxygen in the early stage reduces the number of available nuclei for the later stages. As a result, the final solidification stage has fewer graphite nuclei, less graphite precipitation, and weaker expansion. The last liquid regions then solidify without sufficient self-feeding, and dispersed shrinkage forms in the ductile iron castings.
I also considered the possibility that the molten iron was under-inoculated. In hypoeutectic iron, insufficient inoculation can also cause high recalescence because the melt lacks sufficient nuclei and solidifies at a higher undercooling before recalescence. However, the thermal analysis showed that the iron was eutectic and had no primary austenite. Under this condition, over-inoculation is a more likely cause of excessive early graphite precipitation. Therefore, I decided to reduce the amount of inoculant added to the pouring ladle and to observe the effect on the thermal analysis curve and on the shrinkage defects in the ductile iron castings.
4. Optimization of the Inoculation Process
For the next ladle, I reduced the amount of silicon-calcium-barium inoculant added to the pouring ladle from 0.4% to 0.3%. The other process parameters were kept unchanged. The nodularization treatment, the stream inoculation of 0.1%, the carbon equivalent, the pouring temperature, and the pouring time were the same as before. After the nodularized iron was transferred to the pouring ladle, I took a sample from the top of the ladle and poured it into a quick thermal analysis cup containing 0.1% silicon-calcium-barium inoculant with a particle size of 0.2–0.7 mm. The thermal analysis result was significantly different.
The recalescence temperature decreased from 9.2 °C to 6.5 °C. The thermal conductivity coefficient increased from 20 to 24. The reduction in recalescence was 2.7 °C, and the increase in the thermal conductivity coefficient was 4 units. This indicated that the early graphite nucleation and precipitation were reduced, while the late graphite nucleation and precipitation were improved. In other words, the graphite expansion became more balanced. The early expansion was no longer excessive, so the feeding channels remained open longer. The late expansion was stronger, so the last liquid regions could self-feed more effectively. This change was favorable for reducing dispersed shrinkage in the ductile iron castings.
| Parameter | Before Optimization | After Optimization | Change | Effect on Ductile Iron Castings |
|---|---|---|---|---|
| Ladle inoculant addition | 0.4% | 0.3% | −0.1% | Reduced early graphite nucleation |
| Stream inoculant addition | 0.1% | 0.1% | 0 | Maintained surface and edge inoculation |
| Recalescence temperature | 9.2 °C | 6.5 °C | −2.7 °C | Less early expansion |
| Thermal conductivity coefficient | 20 | 24 | +4 | More late graphite expansion |
| Solidification type | Eutectic | Eutectic | Unchanged | Low macro-shrinkage risk |
| Expected dispersed shrinkage | High | Low | Improved | Better self-feeding |
I used this ladle of molten iron to produce wheel hub castings. After machining, I did not find any shrinkage cavities or dispersed shrinkage in the castings. I then repeated the optimized inoculation process in large-scale production. The process was stable, and the dispersed shrinkage rejection rate decreased significantly. In one production batch, 2,716 castings were poured, and 23 additional castings from a previous missed inspection were also included in the inspection. The total number of inspected castings was 2,739. The number of castings rejected due to dispersed shrinkage was 24. This corresponds to a dispersed shrinkage rejection rate of about 0.88%. The overall rejection rate, including sand holes, gas holes, damage, and other defects, was about 3.91%. The company requirement was to keep the dispersed shrinkage rejection rate below 2% and the overall rejection rate after machining below 5%. The optimized process met these requirements.
| Item | Value | Unit |
|---|---|---|
| Product type | 2519 wheel hub | — |
| Single casting weight | 18.68 | kg |
| Poured quantity | 2,716 | pieces |
| Poured tonnage | 50.73 | t |
| Inspected quantity | 2,739 | pieces |
| Inspected tonnage | 51.16 | t |
| Accepted quantity | 2,714 | pieces |
| Accepted tonnage | 50.70 | t |
| First-time scrap quantity | 25 | pieces |
| First-time scrap rate | 0.91 | % |
| Second-time scrap quantity | 82 | pieces |
| Second-time scrap rate | 2.99 | % |
| Total scrap quantity | 107 | pieces |
| Overall rejection rate | 3.91 | % |
| Dispersed shrinkage scrap quantity | 24 | pieces |
| Dispersed shrinkage rejection rate | 0.88 | % |
The scrap categories in this batch included sand holes, damage, sand burning, cold shut, black skin, unclear marking, slag holes, missing core, cracks, poor sand core, gas holes, core shifting, surface peeling, excess material, dispersed shrinkage, sand squeezing, sand dropping, shrinkage depression, air leakage, misrun, poor material, edge breakage, grinding scrap, blocked water passage, corner breakage, core breakage, scraping, pattern draw problems, hard material, deformation, shrinkage cavity, mold expansion, unqualified sonic test, and other defects. Among these, dispersed shrinkage was the defect most directly related to the thermal analysis and inoculation optimization. The table below summarizes the main scrap categories and their contribution.
| Scrap Category | Quantity | Typical Cause | Relation to Thermal Analysis |
|---|---|---|---|
| Dispersed shrinkage | 24 | Insufficient late graphite expansion | Directly controlled by inoculation and thermal analysis |
| Sand holes | 28 | Sand core or mold erosion | Not directly related |
| Damage | 29 | Handling and cleaning | Not directly related |
| Gas holes | 1 | Gas entrapment | Indirectly related to pouring temperature |
| Cold shut | — | Low pouring temperature | Indirectly related to pouring practice |
| Other defects | 25 | Various | Mostly not related |
| Total | 107 | — | — |
5. Discussion of the Thermal Analysis Mechanism
The results showed that thermal analysis is a powerful tool for controlling dispersed shrinkage in ductile iron castings. The cooling curve of the molten iron contains information about nucleation, growth, recalescence, and final solidification. By comparing the curve shape and the key parameters before and after reducing the inoculant addition, I was able to identify over-inoculation as the root cause of the excessive early graphite expansion and insufficient late graphite expansion. This was not obvious from chemistry alone, because the carbon and silicon contents were within the normal range. It was also not obvious from simulation alone, because the simulation could not accurately describe the stochastic nucleation and growth of graphite in different inoculation conditions. Thermal analysis provided a direct measurement of the solidification behavior of the ductile iron castings.
In ductile iron castings, the volume change during solidification can be expressed conceptually as the sum of liquid contraction, austenite contraction, and graphite expansion. The balance can be written as:
$$ \Delta V_{total} = \Delta V_{liquid} + \Delta V_{austenite} + \Delta V_{graphite} $$
For self-feeding to occur, the graphite expansion must compensate for the liquid and austenite contraction at the right time. If graphite expansion occurs too early, it can cause a pressure increase in the early stage and can close the feeding channels. If it occurs too late, it cannot compensate for the contraction of the last liquid regions. The ideal condition is a moderate recalescence and a high thermal conductivity coefficient at the end of solidification. The recalescence can be defined as:
$$ \Delta T_r = T_{max} – T_{min} $$
where \( T_{max} \) is the maximum eutectic temperature after recalescence and \( T_{min} \) is the minimum eutectic temperature before recalescence. A high \( \Delta T_r \) indicates a large amount of latent heat released in a short time, which corresponds to rapid early graphite growth. The thermal conductivity coefficient is an indirect measure of the final stage of graphite precipitation. A low value indicates that the final liquid regions do not have enough graphite expansion to self-feed. In my case, the initial \( \Delta T_r \) was 9.2 °C and the thermal conductivity coefficient was 20. After reducing the ladle inoculant from 0.4% to 0.3%, \( \Delta T_r \) decreased to 6.5 °C and the thermal conductivity coefficient increased to 24. This shift was exactly what was needed to reduce dispersed shrinkage in the ductile iron castings.
| Parameter | Definition | Target for Ductile Iron Castings | Risk If Too High | Risk If Too Low |
|---|---|---|---|---|
| Recalescence temperature, \( \Delta T_r \) | \( T_{max} – T_{min} \) | 2–7 °C | Excessive early expansion, feeding channel closure | Insufficient nucleation, high undercooling, chill risk |
| Thermal conductivity coefficient | Indicator of final graphite precipitation | 22–28 | Excessive late expansion, possible graphite floating | Insufficient late expansion, dispersed shrinkage |
| Solidification type | Primary phase before eutectic | Eutectic or slightly hypoeutectic | Hypereutectic, graphite floating | Hypoeutectic, macro-shrinkage |
| Carbon equivalent | CE = C + Si/3 | 4.33–4.47% | Graphite floating, coarse graphite | Chill, shrinkage cavity |
The carbon equivalent is an important parameter for ductile iron castings. In my process, the final carbon equivalent was 4.33%–4.47%. This range is close to the eutectic point, which favors self-feeding. The formula for carbon equivalent is:
$$ CE = C + \frac{Si}{3} $$
where \( C \) and \( Si \) are the carbon and silicon contents in weight percent. For the final composition of 3.50%–3.60% C and 2.50%–2.60% Si, the carbon equivalent was:
$$ CE = 3.55 + \frac{2.55}{3} = 4.40\% $$
This value is within the target range. The eutectic solidification observed in thermal analysis confirmed that the carbon equivalent was well controlled. Therefore, the main problem was not the carbon equivalent but the inoculation practice. This is an important lesson for ductile iron castings: a good chemistry range does not guarantee good solidification behavior. The nucleation state of the melt must also be controlled.
6. Practical Rules for Using Thermal Analysis in Production
Based on this work, I established several practical rules for using thermal analysis to control ductile iron castings. These rules are summarized below.
- Take the thermal analysis sample from the pouring ladle at the same location and time for every ladle. The top of the ladle is preferred because it represents the iron that will enter the mold cavity last.
- Use a quick thermal analysis cup that contains the same stream inoculant as the production line. This makes the sample cup condition similar to the mold filling condition.
- Record the recalescence temperature and the thermal conductivity coefficient for every ladle. Do not rely only on carbon and silicon contents.
- Set a target recalescence range of 2–7 °C for eutectic ductile iron castings. If the recalescence is above 7 °C, consider reducing the ladle inoculant addition or checking the nodularization process.
- Set a target thermal conductivity coefficient of 22–28. If the value is below 22, the final solidification stage may lack graphite expansion, and dispersed shrinkage may occur.
- Adjust the inoculant addition in small steps, such as 0.05% or 0.1%, and verify each change with thermal analysis before applying it to mass production.
- Keep the time from nodularization to pouring as short as possible. In my process, it was controlled within 9 minutes. A longer holding time can cause inoculation fade and can change the solidification curve.
- Control the pouring temperature within a narrow range, such as 1,350–1,400 °C. A temperature that is too high can increase shrinkage, while a temperature that is too low can cause cold shut and misrun.
- Use simulation as a design tool, but validate the final process with thermal analysis and production trials. Simulation cannot fully capture the inoculation state of the molten iron.
| Control Item | Target or Action | Reason for Ductile Iron Castings |
|---|---|---|
| Sample location | Top of pouring ladle | Represents the last iron entering the mold |
| Sample cup inoculant | 0.1% stream inoculant | Simulates production stream inoculation |
| Recalescence temperature | 2–7 °C | Avoid excessive early graphite expansion |
| Thermal conductivity coefficient | 22–28 | Ensure sufficient late graphite expansion |
| Ladle inoculant adjustment | Reduce by 0.1% if recalescence is high | Balance early and late graphite precipitation |
| Time to pouring | ≤9 min | Minimize inoculation fade |
| Pouring temperature | 1,350–1,400 °C | Balance fluidity and shrinkage |
| Carbon equivalent | 4.33–4.47% | Promote eutectic solidification and self-feeding |
7. Economic and Quality Benefits
The optimization of the inoculation process had a direct effect on the rejection rate of the ductile iron castings. Before optimization, the dispersed shrinkage rejection rate in mass production was above 12%. After optimization, it decreased to about 0.88% in the reported batch. The overall rejection rate was about 3.91%, which was below the company target of 5%. This reduction not only saved material and machining costs but also improved delivery reliability and customer satisfaction. In addition, reducing the ladle inoculant addition from 0.4% to 0.3% reduced the consumption of inoculant by 25%. This is a direct cost saving. It also reduced the risk of over-inoculation, which can cause graphite degeneration, excessive slag, and other defects in ductile iron castings.
| Metric | Before Optimization | After Optimization | Improvement |
|---|---|---|---|
| Dispersed shrinkage rejection rate | >12% | ≈0.88% | Reduced by more than 11 percentage points |
| Overall rejection rate | Not reported | 3.91% | Below 5% target |
| Ladle inoculant addition | 0.4% | 0.3% | 25% reduction |
| Recalescence temperature | 9.2 °C | 6.5 °C | Within 2–7 °C target |
| Thermal conductivity coefficient | 20 | 24 | Within 22–28 target |
| Process stability | Poor | Good | Stable mass production |
8. Key Equations and Their Practical Use
For convenience, I summarize the key equations used in this work. These equations can be applied to other ductile iron castings when designing gating and feeding systems and when interpreting thermal analysis results.
| Equation Name | Expression | Use |
|---|---|---|
| Quality perimeter quotient | $$ Q_m = \frac{G}{M_c^3} $$ | Evaluate the feeding difficulty of a casting |
| Shrinkage time fraction | $$ P_c = \frac{1.0}{e^{(0.5M_c + 0.01Q_m)}} $$ | Estimate the time available for shrinkage |
| Shrinkage modulus coefficient | $$ f_2 = \sqrt{P_c} $$ | Calculate the casting shrinkage modulus |
| Casting shrinkage modulus | $$ M_s = f_2 M_c $$ | Determine the required feeder modulus |
| Feeder modulus | $$ M_R = f_1 f_3 M_s $$ | Design the feeder size |
| Recalescence temperature | $$ \Delta T_r = T_{max} – T_{min} $$ | Assess early graphite expansion |
| Carbon equivalent | $$ CE = C + \frac{Si}{3} $$ | Control eutectic solidification |
| Volume change balance | $$ \Delta V_{total} = \Delta V_{liquid} + \Delta V_{austenite} + \Delta V_{graphite} $$ | Understand self-feeding in ductile iron castings |
9. Conclusions
1. I found that improving the gating and feeding system alone could not eliminate dispersed shrinkage in the ductile iron castings. The gating and feeding design based on balanced solidification and modulus calculation was useful, but it did not control the solidification behavior of the molten iron after inoculation.
2. Thermal analysis of the molten iron before pouring revealed that the iron after nodularization and inoculation had excessive early graphite expansion and insufficient late graphite expansion. The recalescence temperature was 9.2 °C, which was above the ideal range of 2–7 °C, and the thermal conductivity coefficient was 20, which was below the ideal range of 22–28.
3. I identified over-inoculation as the cause of the excessive early graphite precipitation. The ladle inoculant addition of 0.4% was too high for the eutectic ductile iron used in these wheel hub castings.
4. By reducing the ladle inoculant addition from 0.4% to 0.3%, I decreased the recalescence temperature to 6.5 °C and increased the thermal conductivity coefficient to 24. This produced a more balanced graphite precipitation sequence.
5. The optimized inoculation process reduced the dispersed shrinkage rejection rate of the ductile iron castings from more than 12% to less than 2%. In one mass production batch, the dispersed shrinkage rejection rate was about 0.88%, and the overall rejection rate was about 3.91%.
6. Thermal analysis is an effective tool for diagnosing and controlling the solidification quality of ductile iron castings. It can detect over-inoculation and under-inoculation, and it provides a quantitative basis for adjusting the inoculation process. I recommend using thermal analysis together with chemical analysis, simulation, and production trials to achieve stable, defect-free ductile iron castings.
7. For future production, I will continue to monitor the recalescence temperature and the thermal conductivity coefficient for every ladle of ductile iron. I will keep the ladle inoculant addition within the range that produces a recalescence of 2–7 °C and a thermal conductivity coefficient of 22–28. This will help maintain a low rejection rate and ensure the internal quality of safety-critical ductile iron castings.