In my daily work, I have learned that the production of a heavy-section ductile iron casting is fundamentally different from the production of a thin-walled ductile iron casting. The word “ductile iron casting” often suggests a material that contains spherical graphite and therefore possesses high strength, good elongation and reliable impact resistance. Yet when the wall thickness of a ductile iron casting increases, the local solidification time becomes much longer, and the metallurgical stability of the melt becomes much harder to maintain. The core of the casting may fail to meet the requirements of the specification even if the surface and the attached test coupons look completely acceptable. This problem is especially severe in large wind-power components, which often contain heavy sections where the molten metal has to remain liquid for a long time before final solidification.
In the present study, I focused on a real wind-power hub that had a total pouring weight of about 15000 kg. The principal wall thickness of this hub was 90 mm, the maximum wall thickness was 150 mm, and the minimum wall thickness was 40 mm. When I examined a full section of the hub, the heavy part between 100 mm and 150 mm wall thickness showed a very poor graphite structure. The nodularity in that heavy section was only between 40% and 75%, and a large amount of chunky graphite had formed in the center of the section. The most alarming observation was that the attached test coupons from the same castings satisfied all conventional acceptance criteria. The attached coupon indicated that the ductile iron casting was acceptable, but the actual body of the hub still contained abnormal graphite. This is a classic example of the danger of relying only on attached coupons for the qualification of a heavy-section ductile iron casting.
For a heavy-section ductile iron casting, the graphite morphology is not uniform from the surface to the center. With increasing section thickness, the solidification time increases, spheroidization fades, inoculation fades, and anti-spheroidizing trace elements become enriched in the residual liquid. The graphite shape in the casting changes from spherical graphite to clustered graphite, then to vermicular graphite, and finally to exploded graphite or chunky graphite in the center. This phenomenon is often called the “sandwich effect” because the cross section appears to have different zones. In addition, I have observed aligned graphite rows and graphite segregation in thick sections. Large graphite nodules may grow at the center of a heavy-section ductile iron casting, and some graphite nodules develop long tails or hairy surfaces. These abnormal graphite structures reduce the local nodularity and seriously reduce the mechanical properties of the ductile iron casting, especially the elongation and dynamic toughness.
The formation of chunky graphite is usually associated with the rupture of the austenite shell. In my understanding, during solidification of a heavy-section ductile iron casting, convective flow and thermal currents can break the austenite shell that surrounds a growing graphite nodule. When the shell is damaged, the graphite is no longer constrained by the austenite diffusion barrier. Carbon can then be transported directly from the liquid to the growing graphite surface. This permits rapid and unstable growth. Elements such as calcium, strontium and cerium can also segregate at the austenite shell and change the interfacial energy. As a result, the graphite growth mode changes from spherical to chunky or branched. The core of the chunky graphite is still nodular, but the outer portion branches severely. This kind of graphite is a degenerate form of graphite and must be controlled by process design rather than by inspection alone.
Motivation for the experiment
In order to improve the nodularity of the heavy section of a large ductile iron casting, I decided to simulate the heavy-section condition using a large test block. The test block was designed to reproduce the cooling rate and the solidification time of the thickest parts of the production hub. The goal was to compare different spheroidization and inoculation methods. I also wanted to test the influence of carbon content and the addition of antimony. The final objective was to find a process that could reliably deliver a nodularity of at least 85% at the center of a heavy-section ductile iron casting while maintaining acceptable ferrite content and impact toughness.
I used a 500 mm × 500 mm × 500 mm cube-shaped test block. This block has a very large modulus, and therefore a very long solidification time. The pattern was made with a draft angle of 5° in the direction of increasing wall thickness. I adopted a bottom-gated pouring system with a sprue diameter of 60 mm. At the top center of the test block, I placed an insulating riser with a diameter of 180 mm. This arrangement was intended to mimic the feeding behavior and the thermal gradient of the hub. In such a heavy-section ductile iron casting, the center of the block remains hot for a long period, and the temperature gradient is low. This is the most difficult condition for obtaining a fully nodular structure.
The metal charge consisted of 80% high-purity pig iron, 10% carbon steel scrap and 10% return scrap from wind-power ductile iron castings. I controlled the carbon equivalent in the range from 4.2 to 4.4. The melt was produced in a 3 ton induction furnace. The tapping temperature was between 1490°C and 1510°C, and the pouring temperature was between 1370°C and 1380°C. The total pouring time was controlled to be less than 15 minutes. This short pouring time is important because any delay between nodulization and pouring would increase the risk of spheroidization fading.
Five experimental process schemes
I designed five different process schemes in order to identify the most effective method for improving the center nodularity. All five schemes used the same base sulfur-oxygen inoculant, 0.15%, added immediately before pouring. The schemes differed in the type of nodulizer, the carbon content, and the use of antimony. The high-calcium-barium alloy was used in two ways. First, 0.1% high-calcium-barium was used as a covering material over the nodulizer. Second, 0.4% high-calcium-barium was added into the stream during ladle filling. This high-calcium-barium addition helps to stabilise the spheroidizing reaction and also improves inoculation. The target chemical compositions of the five schemes are summarized in Table 1.
| Scheme | C (%) | Si (%) | Mn (%) | P (%) | S (%) | Special addition |
|---|---|---|---|---|---|---|
| 1 | 3.92 | 0.80 | 0.10 | 0.002 | 0.012 | — |
| 2 | 3.85 | 0.85 | 0.12 | 0.0021 | 0.013 | 0.005–0.006 Sb |
| 3 | 3.86 | 0.81 | 0.11 | 0.023 | 0.014 | Y-based heavy rare earth |
| 4 | 3.83 | 0.82 | 0.12 | 0.024 | 0.012 | 0.9 mixed RE + 0.3 Y-RE |
| 5 | 3.84 | 0.83 | 0.14 | 0.002 | 0.013 | 0.005 Sb |
Scheme 1 represented the normal production process for the wind-power hub. It used 1.2% mixed rare-earth nodulizer, 0.1% high-calcium-barium as a covering material, 0.4% high-calcium-barium added during teeming, and 0.15% sulfur-oxygen inoculant. Scheme 2 was the same as Scheme 1 except that I reduced the carbon content from the normal range of 3.85% to 3.95% down to 3.80% to 3.90%. In addition, I placed 0.006% antimony in a separate non-spheroidizing pit in the treatment ladle. This was done to delay the reaction of antimony and to avoid direct contact with the nodulizer. Scheme 3 used 1.2% yttrium-based heavy rare-earth nodulizer, with the same covering and teeming additions as Scheme 1. Scheme 4 used a combination of 0.9% mixed rare-earth nodulizer and 0.3% yttrium-based heavy rare-earth nodulizer. Scheme 5 used 1.0% mixed rare-earth nodulizer, 0.2% yttrium-based heavy rare-earth nodulizer, and 0.005% antimony. The treatment and inoculation procedure for Scheme 5 was otherwise identical to that used in the normal process.
The antimony addition in Schemes 2 and 5 deserves special attention because antimony is known to counteract the formation of chunky graphite. In a heavy-section ductile iron casting, residual cerium and other rare-earth elements segregate at the graphite–liquid interface and promote branching. Antimony, because of its strong surface-active behavior, can partly neutralize that effect. However, antimony must be carefully controlled because it is also a pearlite stabilizing element. If too much antimony is added, the matrix may become pearlitic, and the impact property of the ductile iron casting may decrease. In my experiments, therefore, I kept the antimony addition at the low level of 0.005% in Scheme 5 and 0.005% to 0.006% in Scheme 2.
Carbon equivalent and solidification parameters
The carbon equivalent of a ductile iron casting can be calculated from the carbon, silicon and phosphorus contents. In the foundry, I have often used the simple formula:
$$
CE = w(\mathrm{C}) + \frac{1}{3}\left[w(\mathrm{Si}) + w(\mathrm{P})\right]
$$
For Scheme 1, the target chemical composition gives:
$$
CE_1 = 3.92 + \frac{0.80 + 0.002}{3} = 4.19
$$
For Scheme 5, the target composition gives:
$$
CE_5 = 3.84 + \frac{0.83 + 0.002}{3} = 4.12
$$
The reduction of carbon equivalent from about 4.19 to about 4.12 is small but important. A lower carbon equivalent reduces the amount of excess carbon available for graphite flotation, exploded graphite and carbon segregation. In a heavy-section ductile iron casting, a high carbon equivalent tends to increase the formation of large, irregular graphite at the center. By lowering the carbon content, I was able to reduce the tendency for exploded graphite.
Solidification time is closely related to the modulus of the casting. According to Chvorinov’s rule, the solidification time can be expressed as:
$$
t_s = B \left( \frac{V}{A} \right)^2
$$
where \(V\) is the volume, \(A\) is the cooling surface area, and \(B\) is a constant that depends on the mold material and the metal properties. For a 500 mm cube, the modulus is:
$$
M = \frac{V}{A} = \frac{a^3}{6a^2} = \frac{500}{6} \approx 83.3 \, \mathrm{mm}
$$
This modulus is much larger than the modulus of a conventional attached test coupon. Therefore, the solidification time of the test block center is much longer. During this long solidification time, spheroidization and inoculation can fade, and rare-earth elements can segregate. This is why the center of the test block is a very severe location for a heavy-section ductile iron casting.
Graphite growth in the heavy section is controlled by the diffusion of carbon through the austenite shell. The carbon flux can be written as:
$$
J_C = D_C \frac{C_{\gamma/\mathrm{gr}} – C_{\gamma/L}}{\Delta r}
$$
where \(D_C\) is the diffusion coefficient of carbon in austenite, \(C_{\gamma/\mathrm{gr}}\) is the carbon concentration in austenite at the graphite interface, \(C_{\gamma/L}\) is the carbon concentration in austenite at the liquid interface, and \(\Delta r\) is the thickness of the austenite shell. When the austenite shell remains intact, the growth rate of graphite is controlled by solid-state diffusion, which is relatively slow. When the shell is broken, liquid carbon can reach the graphite surface directly. The growth rate increases dramatically, and the graphite develops branches. This is the fundamental reason why chunky graphite forms in the center of a heavy-section ductile iron casting.
Sampling and examination methods
After pouring and solidification, the test block was sectioned. I cut a slice from the central vertical plane with a thickness of 70 mm. Metallographic specimens were taken from the center of this slice and from the side-wall region. The side-wall region represented a faster-cooling area, while the center of the slice represented the region with the longest solidification time. I examined the graphite morphology using an optical microscope and measured the nodularity using the same standard method that I use for production evaluation. I counted the graphite particles and calculated the percentage of type V and type VI graphite, which are the acceptable graphite types in the relevant cast-iron standard. The nodularity value reported in my tables therefore represents the combined proportion \(N_{\mathrm{V+VI}}\).
I also measured the final chemical composition of each test block. The final compositions are given in Table 2. The final carbon contents are slightly lower than the target values because some carbon is lost during nodulization and during the holding time before pouring. The silicon contents are higher because of the inoculants and nodulizers. The residual antimony in Scheme 2 and Scheme 5 was about 0.005%, which is in the expected range.
| Scheme | C (%) | Si (%) | Mn (%) | P (%) | S (%) | Sb (%) |
|---|---|---|---|---|---|---|
| 1 | 3.75 | 1.75 | 0.18 | 0.019 | 0.008 | 0.001 |
| 2 | 3.62 | 1.77 | 0.17 | 0.020 | 0.009 | 0.005 |
| 3 | 3.68 | 1.80 | 0.20 | 0.018 | 0.008 | 0.001 |
| 4 | 3.67 | 1.82 | 0.21 | 0.020 | 0.007 | 0.001 |
| 5 | 3.66 | 1.84 | 0.19 | 0.022 | 0.006 | 0.005 |
Metallographic results of the test blocks
The metallographic results showed a large difference between the normal process and the optimized process. In Scheme 1, which represented the normal production process, the center of the 500 mm test block had a nodularity of only 35%. This is far below the acceptance limit that is usually applied to a heavy-section ductile iron casting. The graphite size grade in the center was 4 to 5. The graphite morphology was irregular, with large exploded graphite particles and a significant amount of chunky graphite. This confirmed my suspicion that the attached coupon on the production hub did not represent the actual heavy-section center. In a thin area near the surface, the graphite might look acceptable, but the center of the section could still contain degenerate graphite.
Scheme 2 produced a clear improvement. The reduction of carbon content and the addition of 0.006% antimony improved the center nodularity to 75%. The graphite size grade was 5. The exploded graphite was less frequent, and the chunky graphite was reduced. However, the nodularity was still below 85%, and there were still signs of spheroidization fading and inoculation fading in the center. The very long solidification time of the 500 mm block was more than the normal mixed rare-earth nodulizer could tolerate. Thus, although antimony and lower carbon were helpful, they were not sufficient by themselves.
Scheme 3 used the yttrium-based heavy rare-earth nodulizer. This produced a center nodularity of 85%. The yttrium-based nodulizer had better resistance to fading, and the graphite remained nodular for a longer time. However, I observed two problems. First, the graphite nodules were large and not as round as those produced by the mixed rare-earth nodulizer. Second, the segregation of yttrium and other rare-earth elements at the end of solidification generated a small amount of pearlite in the matrix. This pearlite is unfavourable for low-temperature impact toughness, especially in a wind-power ductile iron casting that must operate under cold conditions. Therefore, the use of yttrium-based heavy rare-earth nodulizer alone was not completely satisfactory.
Scheme 4 combined 0.9% mixed rare-earth nodulizer with 0.3% yttrium-based heavy rare-earth nodulizer. The center nodularity reached 87%. This was better than the normal process and better than the simple lower-carbon plus antimony process. The combination of light and heavy rare earth elements provided both early spheroidization and long-term protection. Nevertheless, the center still contained a small amount of chunky graphite and a small amount of pearlite. The chunky graphite was not enough to reduce the nodularity below 85%, but I wanted a more robust process margin for production.
Scheme 5 was the most successful process. It used 1.0% mixed rare-earth nodulizer, 0.2% yttrium-based heavy rare-earth nodulizer, and 0.005% antimony. The center nodularity of the 500 mm test block reached 90%. The graphite size grade was 5. The graphite morphology was smooth, round and uniformly distributed. I did not observe significant chunky graphite in the center. The matrix was essentially ferritic, and the amount of pearlite was very small. This result showed that a carefully combined spheroidizing system can overcome the long solidification time that is typical of a heavy-section ductile iron casting. The metallographic results are summarized in Table 3.
| Specimen location | Nodularity (V+VI) | Graphite size grade | Remarks |
|---|---|---|---|
| Scheme 1, center | 35% | 4–5 | Chunky graphite, exploded graphite |
| Scheme 2, center | 75% | 5 | Some fading, reduced chunky graphite |
| Scheme 3, center | 85% | 4–5 | Large nodules, some pearlite |
| Scheme 4, center | 87% | 5 | Slight chunky graphite, slight pearlite |
| Scheme 5, center | 90% | 5 | Round graphite, mainly ferritic matrix |
I believe that Scheme 5 worked because it used three mechanisms at the same time. First, the mixed rare-earth nodulizer provided a strong and immediate spheroidizing effect. Second, the yttrium-based heavy rare-earth nodulizer acted as a reservoir of nodulizing elements that could be released late in solidification, preventing spheroidization fading. Third, the antimony suppressed the branching of graphite and reduced the formation of chunky graphite. In addition, the slightly lower carbon content reduced the danger of carbon excess and exploded graphite. Together, these effects allowed the center of the heavy-section ductile iron casting to maintain a nodularity of 90%.
Effects of carbon content
The carbon content has a very important influence on the graphite structure of a heavy-section ductile iron casting. When the carbon equivalent is too high, the melt contains a large amount of graphite that can grow freely. In the early stages of solidification, the graphite may float and accumulate near the riser or at the top of the casting. In the center of a thick section, this excess carbon contributes to the formation of exploded graphite. Exploded graphite appears when the graphite growth front breaks down into many branched fragments. The normal process in my foundry had a carbon range from 3.85% to 3.95%. In Scheme 2, I reduced the carbon range to 3.80% to 3.90%. The final carbon content in Scheme 2 was 3.62%, while the final carbon content in Scheme 1 was 3.75%. This modest reduction was enough to reduce the amount of exploded graphite and to improve the center nodularity from 35% to 75%.
It is important to note that the carbon content cannot be reduced too much in a ductile iron casting. If the carbon equivalent is too low, the feedability of the casting deteriorates, and the risk of shrinkage porosity increases. Therefore, the carbon content must be balanced between the need to avoid graphite degeneration and the need to maintain good feeding quality. In my experiments, Scheme 5 produced a final carbon content of 3.66%, which is low enough to control exploded graphite but still high enough to avoid serious shrinkage problems. The successful production verification confirmed that this carbon content is practical for a wind-power hub.
Effects of antimony
Antimony is a trace element that has a very strong effect on the graphite morphology of a heavy-section ductile iron casting. In a normal ductile iron casting, residual rare-earth elements such as cerium and lanthanum are present because of the nodulizer. These rare-earth elements can segregate at the austenite–graphite interface and promote the transition from spherical graphite to chunky graphite. The chunky graphite is especially dangerous because it is not easily detected by the attached test coupon and yet it destroys the elongation and toughness of the material. Antimony can counteract the harmful effect of rare-earth segregation. I added antimony in Scheme 2 and Scheme 5. The addition of 0.005% to 0.006% antimony suppressed the chunky graphite formation to a large extent. In Scheme 5, the combination of antimony with yttrium-based heavy rare-earth nodulizer was particularly effective because the yttrium protected the spheroidization during the long solidification time, while the antimony prevented the rare-earth elements from destabilizing the graphite interface.
I also considered the possible formation of stable antimony compounds. The reaction between cerium and antimony can be represented in a simplified way:
$$
[\mathrm{Ce}] + [\mathrm{Sb}] \rightarrow \mathrm{CeSb}
$$
Although the actual metallurgy is more complex, this simplification shows that antimony can combine with the harmful rare-earth elements and reduce their ability to cause chunky graphite. At the same time, antimony must be carefully limited because it is a pearlite stabilizer. In my test blocks, Scheme 5 with 0.005% antimony produced a mostly ferritic matrix. The low level of antimony was sufficient to control chunky graphite without causing excessive pearlite. This is why the impact values remained acceptable.
Effects of rare-earth type
The type of rare-earth element used in the nodulizer changes the behaviour of the melt in heavy sections. Mixed rare-earth nodulizer, which contains cerium-rich light rare-earth elements, is very effective for immediate spheroidization. It produces small and round graphite nodules in thin sections. However, in a heavy-section ductile iron casting, the light rare-earth elements can fade during the long solidification time. As the temperature decreases, cerium and lanthanum segregate in the residual liquid and promote the growth of chunky graphite. Therefore, the normal process using only mixed rare-earth nodulizer failed to provide acceptable center nodularity in the test block.
Yttrium-based heavy rare-earth nodulizer behaves differently. The yttrium-containing compounds are more stable than the cerium compounds. They do not fade as quickly, and they can continue to support graphite nucleation late in solidification. This is why Scheme 3 produced a nodularity of 85% even in the center of the 500 mm test block. However, yttrium also has a tendency to segregate strongly in the intercellular regions during the final stage of solidification. This segregation can stabilize pearlite and reduce the ductile iron casting’s impact toughness. In order to obtain the best result, I combined mixed rare earth with yttrium-based heavy rare earth. The mixed rare earth provided the initial strong effect, and the yttrium-based heavy rare earth provided the delayed effect. The combination allowed me to use a smaller amount of each type and to avoid excessive segregation of either element.
The final optimized nodulizing system in Scheme 5 was:
$$
\mathrm{Mixed\ RE\ nodulizer} = 1.0\%
$$
$$
\mathrm{Yttrium-based\ RE\ nodulizer} = 0.2\%
$$
$$
\mathrm{Sb} = 0.005\%
$$
This system produced a center nodularity of 90% and did not create any serious pearlite problem. I consider this to be a very robust process for heavy-section ductile iron casting.
Verification on production wheel hubs
After the successful test-block experiments, I applied Scheme 5 to production wind-power hub castings. I selected four identical hubs. Two hubs were produced using the normal production process, and two were produced using Scheme 5. In each hub, I placed a stepped test sample with dimensions of 250 mm × 400 mm in the inner cavity. The stepped test sample reproduced the slower cooling conditions inside the hub. After the hubs were cast and cleaned, I cut samples from the stepped test blocks and performed tensile testing, hardness testing, impact testing and metallographic examination. The results are summarized in Table 4.
| Casting condition | Rm (MPa) | Rp0.2 (MPa) | A (%) | Hardness (HBW) | AKV (J) | Nodularity (V+VI) | Graphite size | Matrix |
|---|---|---|---|---|---|---|---|---|
| Normal process | 304 | 197 | 5 | 120 | 8, 6, 7 | 55% | 4–5 | Ferrite |
| Normal process | 312 | 211 | 12 | 119 | 10, 9, 10 | 71% | 5 | Ferrite |
| Scheme 5 | 358 | 238 | 25.5 | 120 | 12, 10, 12 | 90% | 5–6 | Ferrite |
| Scheme 5 | 342 | 226 | 19.4 | 125 | 12, 11, 13 | 91% | 5–6 | Ferrite |
These results showed that the normal production process did not reliably meet the required center nodularity. The first normal-process casting had a nodularity of only 55%, with 5% elongation. The second normal-process casting was better, with 71% nodularity and 12% elongation, but both castings were far below the 85% nodularity target required for a heavy-section ductile iron casting. In contrast, both Scheme 5 castings had more than 90% nodularity. The tensile strength increased from about 304 to 312 MPa in the normal process to about 342 to 358 MPa in Scheme 5. The yield strength increased from about 197 to 211 MPa to about 226 to 238 MPa. The elongation increased from 5% to 12% in the normal process to 19.4% to 25.5% in Scheme 5. The impact energy also improved. The hardness remained similar, around 120 to 125 HBW, because the matrix was still ferritic. This confirms that the improvement in nodularity did not come at the expense of hardness or machinability.
The graphite size grade in the Scheme 5 samples was 5 to 6, which is slightly finer than the normal process. The graphite nodules were round and evenly distributed. The matrix was almost completely ferritic. These properties are very desirable for a wind-power component because they give good low-temperature toughness and high ductility. The enhancement in mechanical properties was directly related to the improvement in nodularity. When the center of the heavy section contains more true spheroidal graphite and less chunky graphite, the stress concentration around each graphite particle is reduced, and the material can deform plastically before fracture.
Practical recommendations
Based on the test-block experiments and the production verification, I can offer several practical recommendations for improving the center nodularity of a heavy-section ductile iron casting. First, the carbon content should be slightly lower than the normal range used for thin-walled castings. I recommend a target carbon content in the range from 3.80% to 3.90% before treatment. Second, the nodulizing system should contain both light and heavy rare-earth elements. A mixed rare-earth nodulizer should be used as the primary spheroidizing agent, and a yttrium-based heavy rare-earth nodulizer should be added to prevent fading. Third, a small amount of antimony should be added to suppress chunky graphite. The antimony addition should be kept near 0.005% to avoid producing pearlite. Fourth, the inoculation system must be sufficient for the long solidification time. The combination of high-calcium-barium covering and teeming additions, together with sulfur-oxygen in-stream inoculation, proved to be effective in the production castings.
It is also important to realize that the attached test coupon is not sufficient for the quality control of a heavy-section ductile iron casting. The attached coupon may cool much faster than the actual heavy region, and therefore it may show high nodularity even when the center of the casting contains chunky graphite. For critical components, I recommend using a large stepped test block or a separate test block with a representative modulus. The test block should be cut through the center and examined metallurgically. Only then can the true quality of the heavy-section ductile iron casting be known.
The role of cooling rate
Cooling rate is the central factor in the formation of graphite morphology. In a thin-wall ductile iron casting, the cooling rate is high, and the eutectic solidification time is short. There is little time for spheroidization fading or for rare-earth segregation. The graphite remains spherical. In a heavy section, however, the cooling rate is low. The local temperature remains near the eutectic temperature for a long time. This allows the austenite shell to be affected by fluid flow, solute enrichment and interfacial instability. The graphite can then change from spherical to chunky. This is why I selected the 500 mm cube as the test block. It represented the worst-case cooling condition that the hub could experience. The results confirmed that the normal process was not able to maintain the nodularity at the center of such a block.
The relationship between cooling rate and solidification time can be written in terms of the local solidification time:
$$
t_s = \frac{\Delta T}{\dot{T}}
$$
where \(\Delta T\) is the freezing range and \(\dot{T}\) is the local average cooling rate. In the center of a large block, \(\dot{T}\) is very small. Therefore, \(t_s\) is long. During this long period, the spheroidization elements are distributed between the graphite, the austenite, the liquid and the inclusions. If the residual nodulizing elements are not stable, they can react with sulfur, oxygen and other impurities and lose their effect. The heavy rare-earth elements help because they form more stable compounds than the light rare-earth elements. The combined nodulizing system in Scheme 5 maintained the necessary residual elements for the whole solidification range.
Graphite growth and chunky graphite formation
I have also considered the growth mechanism of chunky graphite in more detail. The growth of a graphite nodule in cast iron usually occurs by the diffusion of carbon through the austenite shell. The carbon concentration gradient across this shell drives the growth. The growth velocity can be approximated by:
$$
v = \frac{dr}{dt} = \frac{D_C}{\rho_\gamma r} \left(C^{\gamma/L} – C^{\gamma/G}\right)
$$
where \(r\) is the radius of the graphite nodule, \(\rho_\gamma\) is the density of austenite, \(D_C\) is the diffusion coefficient of carbon in austenite, \(C^{\gamma/L}\) is the carbon concentration at the austenite/liquid interface and \(C^{\gamma/G}\) is the carbon concentration at the austenite/graphite interface. During normal spherical growth, this equation gives a stable, compact nodule. If the austenite shell ruptures, the liquid carbon can flow directly to the graphite, and the growth mode becomes unstable. The graphite then develops branches. The branched graphite later becomes chunky graphite as the branches coarsen and become interconnected in the last-solidifying liquid.
The presence of rare-earth elements at the interface can alter the interfacial energy and make the growth front less stable. The critical radius for a stable graphite nucleus can be expressed as:
$$
r^* = \frac{2 \sigma T_m}{L \Delta T}
$$
where \(\sigma\) is the interfacial energy, \(T_m\) is the melting temperature, \(L\) is the latent heat and \(\Delta T\) is the undercooling. When the interfacial energy is changed by segregation, the critical radius changes, and the graphite may start to grow in a different crystallographic direction. The addition of antimony in Scheme 5 mitigated this effect by neutralizing the harmful rare-earth segregation. The yttrium-based heavy rare-earth nodulizer provided stable nucleation sites and reduced the undercooling required for graphite nucleation. As a result, the graphite remained spherical even in the center of the test block.
Importance of matrix structure
In a heavy-section ductile iron casting, the matrix structure is as important as the graphite morphology. The matrix of a normal unalloyed ductile iron casting is usually ferritic if the cooling rate is slow. Ferritic matrix gives good elongation and toughness. However, if the rare-earth content is too high, or if antimony is added in excess, the matrix can become pearlitic. Pearlite increases the strength but reduces the elongation and low-temperature impact energy. In my optimized Scheme 5, the final composition did not contain excessive rare-earth elements. The antimony content was only 0.005%. Therefore, the matrix remained ferritic. The hardness was still about 120 to 125 HBW, and the impact values were good.
I also observed that in Scheme 3, the use of yttrium-based heavy rare-earth nodulizer alone created a small amount of pearlite in the cooled regions. This pearlite was probably caused by rare-earth segregation in the intercellular regions. The rare-earth-rich carbides can promote the formation of pearlite by locally reducing the amount of graphite carbon and by acting as pearlite nuclei. In the wind-power hub, such pearlite would reduce the low-temperature impact toughness. Therefore, I recommend against using yttrium-based heavy rare-earth nodulizer alone for components that require high impact toughness. The combination of light and heavy rare earth is safer because the amount of yttrium is smaller and the segregation effect is weaker.
Comparison of test-block and production results
The 500 mm test-block results correlated well with the production verification. Scheme 1 produced a test-block center nodularity of 35% and production verification nodularity values of 55% and 71%. The production stepped test block was somewhat smaller and cooled slightly faster than the 500 mm cube, so the nodularity was higher than in the test block. However, the trend was the same. The normal process did not reliably meet the 85% nodularity requirement. Scheme 5 produced a test-block center nodularity of 90% and production verification nodularity values of 90% and 91%. This excellent correlation gives me confidence that the 500 mm test block is a reliable tool for developing new processes for heavy-section ductile iron casting.
The production verification also demonstrated that the improvement in nodularity was associated with a significant improvement in mechanical properties. The tensile strength of Scheme 5 was about 15% higher than the normal process, and the elongation was much higher. This is very important for wind-power components, which are subjected to cyclic loading and low temperature conditions. A heavy-section ductile iron casting with 90% nodularity and a fully ferritic matrix will have better fatigue resistance than a casting with 55% nodularity and chunky graphite. Even if the attached coupon passes the standard inspection, the actual component will be safer and more reliable when the center nodularity is high.
Limitations of attached test coupons
One of the main conclusions from this work is that attached test coupons are not adequate for the qualification of a heavy-section ductile iron casting. The attached coupon is usually designed to cool at a relatively fast rate, and its graphite structure is much finer and more spherical than the center of the actual casting. In my production hubs, the attached coupon passed all requirements, yet the center of the hub had a nodularity as low as 40% to 75%. This is unacceptable for a critical wind-power component. The use of a large test block should therefore be part of the process qualification and periodic verification for any heavy-section ductile iron casting.
I understand that large test blocks increase the cost and complexity of quality assurance. However, the cost of a broken component is much higher than the cost of a test block. For safety-critical castings, the quality must be verified on a section that has the same solidification modulus as the thickest part of the casting. The stepped test block used in my production verification, with dimensions of 250 mm × 400 mm, is a practical compromise. It is smaller than the actual hub but still much more representative than a small attached coupon. The correlation between the 500 mm test block and the stepped test block confirmed that this approach is useful.
Future work
In the future, I plan to investigate the effect of additional trace elements such as bismuth and lead on the graphite morphology of heavy-section ductile iron casting. These elements are known to interact with rare-earth elements and can either promote or suppress chunky graphite. I would also like to study the effect of pouring temperature more systematically. In the present work, the pouring temperature was controlled between 1370°C and 1380°C. This is a common range for large ductile iron castings. A lower pouring temperature could reduce the solidification time and reduce the risk of rare-earth segregation, but it might also create cold-shut and filling problems. The optimum pouring temperature may depend on the geometry of the casting and the gating system.
I also intend to examine the use of a post-inoculation step during pouring, such as in-mould inoculation, to further improve the center structure of a heavy-section ductile iron casting. In-mould inoculation can provide a fresh source of nucleation sites at the moment when the melt enters the mold. This may help to maintain a high graphite nodule count in the slow-cooling center. Combined with the mixed rare-earth and yttrium-based nodulizing system, it could provide an extra margin of safety for very large components.
Conclusions
From the present work, I have reached the following conclusions:
First, a conventional attached test coupon is not representative of the center of a heavy-section ductile iron casting. The normal process used for the wind-power hub produced an acceptable test coupon but only 55% to 71% nodularity in the stepped test block and only 35% nodularity in the 500 mm test block. Therefore, separate large test blocks are necessary for the development and quality control of heavy-section ductile iron castings.
Second, lowering the carbon content from the range of 3.85% to 3.95% down to 3.80% to 3.90% reduces the amount of exploded graphite and improves the center nodularity. In my experiments, this change, together with antimony, improved the nodularity from 35% to 75%.
Third, the addition of 0.005% to 0.006% antimony helps to suppress chunky graphite in a heavy-section ductile iron casting. However, antimony alone cannot completely eliminate chunky graphite, because it does not solve the problem of spheroidization fading.
Fourth, the use of yttrium-based heavy rare-earth nodulizer alone can improve the center nodularity to 85%, but it also produces large graphite nodules and some pearlite, which may reduce the low-temperature impact toughness. Therefore, yttrium-based heavy rare-earth nodulizer is best used in combination with a mixed rare-earth nodulizer.
Fifth, the best result in this study was obtained with a combination of 1.0% mixed rare-earth nodulizer, 0.2% yttrium-based heavy rare-earth nodulizer, and 0.005% antimony. This process produced a center nodularity of 90% in the 500 mm test block and 90% to 91% in the production stepped test block. The mechanical properties improved significantly, with a tensile strength of about 342 to 358 MPa, an elongation of 19.4% to 25.5%, and acceptable impact energy values.
Finally, I believe that the success of a heavy-section ductile iron casting depends not on a single powerful addition but on a balanced combination of carbon control, nodulizer selection, inoculation, trace-element control and test-block verification. The optimized process described in this paper can be used for the production of large wind-power components and other heavy-section ductile iron castings where the attached coupon would otherwise give a false sense of safety.