As a casting engineer who has been deeply involved in the development of large wind power components, I would like to share my practical experience in producing a low-temperature ductile iron main shaft for an ultra-high-power offshore wind turbine. This work is a representative case of modern ductile iron castings used in demanding applications. In this article, I will describe the structural features, technical challenges, casting process design, melting control, and production verification of this component, with a strong emphasis on the key parameters that ensure the reliability and quality of heavy-section ductile iron castings.
The component under discussion is a main shaft for a 10 MW-class offshore wind turbine. It is made of EN-GJS-400-18U-LT ductile iron, with a minimum service temperature of -20 °C. The overall dimensions are 5,280 mm × 5,280 mm × 2,460 mm, and the net weight is 32,672 kg. The casting geometry is characterized by a large flange at the rotor connection side with a maximum wall thickness of 200 mm, while the main cylindrical body has a minimum wall thickness of only 50 mm. Such a significant wall thickness variation, combined with the large external dimensions, makes this component one of the most challenging ductile iron castings in the wind power industry.
The production of high-quality ductile iron castings for low-temperature service requires careful control of metallurgical factors, solidification behavior, and casting defects. I will present a systematic approach that covers the following aspects:
- Structural analysis and technical requirements
- Key difficulties in casting production
- Parting scheme and gating system design
- Feeding and riser system design for heavy sections
- Melting process and chemical composition control
- Nodularization and inoculation treatment
- Process simulation and verification
- Final inspection results and mechanical properties
1. Structural Features and Technical Requirements
The main shaft is a hollow cylindrical component with two large flanges: one for connecting to the rotor hub and the other for the bearing support. The rotor connection flange has a wall thickness of 200 mm, while the bearing flange at the top has a thickness of about 80 mm. The main body wall thickness is only 50 mm. This design creates a very large modulus difference between the heavy flange and the thin cylinder, which often leads to shrinkage porosity if the feeding system is not properly designed. Moreover, the large size of the casting makes it difficult to maintain dimensional accuracy and uniform cooling.
According to the customer specification, the casting must meet the following requirements:
- Mechanical properties according to DIN EN 1563:2005 for EN-GJS-400-18U-LT, tested on separately cast test blocks or attached test blocks.
- Impact toughness at -20 °C: average ≥ 10 J, single value ≥ 7 J for V-notch specimens.
- Ultrasonic testing (UT) according to EN 12680-3:2003, class 2 in critical areas and class 3 in non-critical areas.
- Magnetic particle testing (MT) with fluorescent media according to EN 1369:2012, class 2 in critical areas and class 3 in non-critical areas.
- Metallurgical structure: ferrite matrix with pearlite ≤ 10% for attached test blocks and ≤ 15% for the casting body; graphite form V + VI ≥ 90% for test blocks and ≥ 75% for the casting body; graphite size 4–6 for test blocks.
- Dimensional tolerances according to ISO 8062-3:2007 CT11 for dimensions and CT12 for wall thickness. The final machined weight must be within ±3% of the theoretical weight.
Table 1 summarizes the mechanical property requirements for this low-temperature ductile iron casting.
| Property | Attached Test Block | Casting Body |
|---|---|---|
| Tensile strength / MPa | ≥ 360 | ≥ 310 |
| Yield strength / MPa | ≥ 220 | ≥ 200 |
| Elongation / % | ≥ 12 | ≥ 8 |
| Hardness / HB | 130–180 | 130–180 |
| Impact energy at -20 °C / J (average) | ≥ 10 | – |
| Impact energy at -20 °C / J (single) | ≥ 7 | – |
Table 1. Mechanical property requirements of the ductile iron casting.
The metallurgical requirements are also very strict. Table 2 lists the acceptable values for graphite morphology and matrix structure.
| Category | Matrix Structure | Graphite Form (V+VI) | Graphite Size |
|---|---|---|---|
| Attached test block | Ferrite base, pearlite ≤ 10% | ≥ 90% | 4–6 |
| Casting body | Ferrite base, pearlite ≤ 15% | ≥ 75% | – |
Table 2. Metallurgical structure requirements.
These requirements clearly indicate that the production of such ductile iron castings is not trivial. The combination of low-temperature impact toughness, high internal soundness, and tight dimensional control demands a robust and reproducible manufacturing process.
2. Casting Difficulties
In my experience, there are four major difficulties when making this type of ductile iron castings:
- Shrinkage porosity: Because of the huge difference in wall thickness (200 mm flange vs. 50 mm body), the solidification time varies significantly. The heavy section tends to form isolated liquid pools that cannot be fed by risers if the feeding path is not well designed. Nodular cast iron has a pasty solidification mode with a wide freezing range, which makes it more prone to porosity than gray iron.
- Slag inclusions and surface defects: The requirement of 100% fluorescent MT means that any surface-breaking slag, sand inclusion, or oxide film is unacceptable. Since magnesium-treated ductile iron has a high tendency to form magnesium silicates and other dross, careful metal treatment and gating design are essential.
- Dimensional stability: With a length of more than 5 meters and a weight of over 32 tons, even slight mold deformation or core shift can cause large deviations. The casting must maintain a wall thickness accuracy of CT12, which is very demanding for such a large part.
- Low-temperature toughness: The -20 °C impact requirement of at least 10 J average (and 7 J single) is hard to achieve in heavy sections because slow cooling in thick walls promotes the formation of carbides, pearlite, and non-spheroidal graphite. All of these are detrimental to impact energy.
To overcome these challenges, I adopted a comprehensive engineering approach that combines numerical simulation, optimized gating/riser design, and strict metallurgical process control.
3. Parting Scheme and Mold Assembly
The first step in the casting process design is to determine the orientation and parting lines. For this main shaft, I chose a vertical orientation with the motor coupling flange at the bottom. The parting line was located below the motor coupling flange. The main body of the casting was placed in the cope (upper) part of the mold. A four-box molding method was used, with a separate joint at the rotor connection flange and another joint at the top bearing flange. The pattern was made as a single-piece loose pattern to ensure the outer contour continuity.
The core for the internal cavity was a single large core, which carried the gating system at its bottom. The bottom core print provided positioning, and the core was assembled with minimal handling. This approach reduced the number of core joints, thereby improving dimensional accuracy. Because the outer contour and the critical dimensions were all formed by the same sand body, the consistency of the casting wall thickness was much better than what is typically achieved with multi-piece cores.
The four-box construction allowed the mold to be independently compacted and transported. The schematic layout is as follows:
- Bottom box: supports the motor flange, contains the lower part of the pattern
- Middle box 1: forms the main body and part of the rotor flange
- Middle box 2: continues the main body
- Top box: forms the bearing flange and the upper cavity
This configuration is particularly suitable for large ductile iron castings because it keeps the molding depth manageable and avoids the use of a highly complex core assembly.
4. Gating System Design
To ensure clean and smooth filling of the mold cavity, I designed an unpressurized (open) gating system. The ratio of total cross-sectional areas was set as:
$$ \sum F_{\text{直}}:\sum F_{\text{横}}:\sum F_{\text{内}} = 1:1.8:5 $$
where \(F_{\text{直}}\) is the area of the downsprue, \(F_{\text{横}}\) is the area of the runner, and \(F_{\text{内}}\) is the total area of the ingates. The casting time was set to 160 s, and the flow velocity in the ingates was kept below 0.5 m/s. To achieve this, I used an unpressurized system with a large total gate area. The ingates were arranged evenly around the bottom flange to allow simultaneous filling and to keep the temperature field uniform.
One of the critical aspects of the gating design is the use of ceramic foam filters with a pore density of 10 ppi (pores per inch). These filters not only trap non-metallic inclusions but also act as flow rectifiers. The foam filters were placed in the runner system to remove dross and reduce turbulence. In addition, I used a variable-diameter rectangular ceramic runner that connected to the ingates. This design helped to reduce the local thermal modulus at the ingate contact points, thereby minimizing the risk of shrinkage at those locations.
Another important feature is the slag-removal pouring basin. The pouring basin was designed to hold at least 50% of the total molten metal weight. This allowed the liquid metal to remain in the basin for 30 to 60 seconds before entering the sprue, giving sufficient time for slag and dross to float to the surface. A dam was installed at the bottom of the basin to prevent the dirty top layer from entering the mold cavity. This simple but effective measure plays a significant role in producing clean ductile iron castings.
Figure 1 shows a typical view of the pouring basin and the filtration system used in this project.
I also used MAGMA simulation software to analyze the mold filling process. The simulation results showed that the flow velocity in the ingates remained below 0.5 m/s and that the metal front advanced smoothly without jetting or severe turbulence. This is essential for avoiding the formation of secondary oxide films, which can be detrimental to the mechanical properties of ductile iron castings.
5. Feeding and Riser System Design
As we all know, ductile iron castings solidify in a pasty manner, with a broad solidification interval. The graphite expansion occurs after an austenite shell has formed around the graphite nodules. This expansion can help compensate for shrinkage, but only if the mold is rigid enough. In greensand or weak molds, the expansion may deform the mold, leading to internal porosity. Therefore, a well-designed feeding system is necessary.
For this large main shaft, I arranged two rings of insulated exothermic risers: one ring on the top bearing flange and another ring on the rotor connection flange. The riser design was based on the modulus calculation. The modulus of the riser neck (\(M_{\text{neck}}\)) was set to be 1.1 times the modulus of the casting section (\(M_{\text{casting}}\)), and the modulus of the riser (\(M_{\text{riser}}\)) was set to 1.2 times the modulus of the riser neck. These ratios ensure that the riser solidifies last and can feed the casting during the final stage of solidification.
The feeding equation used is:
$$ M_{\text{neck}} = 1.1 M_{\text{casting}}, \quad M_{\text{riser}} = 1.2 M_{\text{neck}} $$
In addition, the feeding volume of the risers was checked by the following volume balance criterion:
$$ V_{\text{riser}} \ge \frac{\beta V_{\text{casting}}}{\eta} $$
where \(\beta\) is the total volume contraction factor (sum of liquid shrinkage and solidification shrinkage), \(V_{\text{casting}}\) is the volume of the casting section to be fed, and \(\eta\) is the feeding efficiency of the riser. For large ductile iron castings, I typically assume \(\beta\) between 3% and 6% depending on the carbon equivalent and process conditions.
To optimize the solidification sequence, I used chills (cold irons) in limited areas to increase the local cooling rate and to create favorable temperature gradients. The ultimate goal was to ensure that all isolated hot spots are eliminated and that the final solidification occurs near the riser necks. The solidification simulation was used to verify the temperature distribution. The simulation images showed that the temperature gradients remained positive toward the risers, and the last remaining liquid was located within the risers.
6. Melting Process and Chemical Composition
The production of high-quality ductile iron castings starts with the selection of raw materials and the control of chemistry. For low-temperature applications, I placed extra emphasis on limiting trace elements that are harmful to graphite spheroidization and that promote carbide formation. The target chemical composition is given in Table 3.
| Element | Content (wt%) |
|---|---|
| C | 3.6 – 3.9 |
| Si | 1.8 – 2.1 |
| Mn | ≤ 0.2 |
| P | < 0.03 |
| S | < 0.015 |
| Mg | 0.03 – 0.06 |
| Cr, Mo, Cu, Sn, Sb, Pb, Bi, Te, Ti | As low as possible, sum of critical elements < 0.1 |
Table 3. Target chemical composition for -20 °C ductile iron.
Silicon is a key element. It strengthens ferrite by solid solution, but too much silicon raises the ductile-to-brittle transition temperature. In heavy-section ductile iron castings, it is common to keep silicon in the range of 1.8%–2.1%, balancing strength and toughness. Manganese is a strong segregating element that promotes pearlite and carbides at cell boundaries. Therefore, I limited manganese to a maximum of 0.2%, ideally below 0.15%. Phosphorus also increases the transition temperature and should be kept below 0.03%. Sulfur reacts with magnesium, forming sulfide inclusions and consuming nodularizer. The sulfur content in the base iron must be kept below 0.015% to ensure efficient nodularization and minimal dross.
In addition to these major elements, one must pay close attention to trace elements such as lead, antimony, bismuth, tellurium, and titanium. These elements can distort graphite growth, reduce spheroidization, and promote carbide formation. For critical applications, I use a “quality index” based on the sum of individual element equivalents. A common formula used in foundry practice is:
$$ \text{Subversive Index} = 4.4(\text{Ti}) + 2.0(\text{As}) + 2.4(\text{Sb}) + 4.6(\text{Bi}) + 1.0(\text{Pb}) \le 1.0 $$
where the concentrations are in wt%. Keeping this index below 1.0 is beneficial for achieving high nodularity.
Melting was carried out in medium-frequency induction furnaces. The charge consisted of high-purity pig iron, low-sulfur steel scrap, and carbon raiser. The melt was superheated to 1,470–1,490 °C and held for a short time to promote the coalescence and removal of oxide inclusions. The temperature was then adjusted to 1,420–1,440 °C for sampling and final composition adjustment. The aim was to avoid prolonged holding at low temperatures, which can increase the risk of oxidation and nitrogen pickup.
One important parameter for the melt quality is the carbon equivalent, defined as:
$$ \text{CE} = w(\text{C}) + \frac{1}{3} \left( w(\text{Si}) + w(\text{P}) \right) $$
For this casting, I targeted a CE of about 4.3–4.6%. A higher carbon equivalent helps to improve fluidity and promote graphite expansion, but too high a carbon equivalent can lead to graphite flotation in heavy sections. A controlled CE is essential for obtaining the desired microstructure in thick-walled ductile iron castings.
7. Nodularization Treatment
The nodularization treatment was performed using the sandwich method in a treatment ladle. I used a magnesium-rare-earth (RE) nodulizer with a nominal magnesium content of 1.8%. The addition amount was carefully calculated to leave a residual magnesium content of 0.03%–0.06% in the final iron. Too little magnesium results in incomplete spheroidization, while too much magnesium increases dross formation and shrinkage tendency. The treatment ladle had a special chamber to prevent the nodulizer from floating and to ensure a controlled reaction.
The reaction temperature was kept in the range of 1,500–1,520 °C at the beginning of the treatment. After the reaction was complete, the slag was skimmed thoroughly. The iron was then transferred to the pouring ladle. The total time from nodularization to the start of pouring was limited to less than 15 minutes to avoid nodularity decay.
The choice of RE (rare earth) elements in the nodulizer is also important. Small amounts of cerium and lanthanum help to neutralize subversive elements and promote a higher nodule count. However, excessive rare earths can form intermetallic compounds with high melting points, which are difficult to dissolve and can cause defects. The RE content in the nodulizer should be balanced with magnesium.
8. Inoculation Practice
Inoculation is a critical step in producing ductile iron castings with high nodule counts and fully ferritic matrices. For this heavy-section casting, I applied a three-stage compound inoculation process:
- Floating silicon inoculation: A silicon-based inoculant was added to the stream during the nodularization treatment. This step provides a base level of nucleation.
- Ladle inoculation (pour-over): Just before pouring, additional fine-grained inoculant was placed in the pouring ladle. This was the “pour-over” or “ladle stream” inoculation.
- Mold stream inoculation: During the actual pouring, granulated inoculant was added via a hopper attached to the pouring basin. This is the most efficient method, as it provides a large number of potent nuclei just before solidification.
The total inoculation amount was about 0.4%–0.6% of the metal weight, depending on the section thickness and the initial sulfur content. The key is to achieve a high nodule count (>100 nodules/mm²) in the thin sections and >50 nodules/mm² in the thick sections. A higher nodule count reduces the tendency for carbide formation and helps to obtain a fully ferritic matrix.
In heavy sections, the inoculation effect tends to fade quickly. Therefore, the mold stream inoculation must be timed carefully. I used a proprietary long-acting inoculant containing bismuth and rare earths to improve resistance to fade.
9. Pouring and Temperature Control
The pouring temperature was controlled at 1,340–1,360 °C. This temperature range provides sufficient fluidity for filling the complex mold while avoiding excessive shrinkage during solidification. For thick-section ductile iron castings, a lower pouring temperature is generally preferred because it reduces the risk of hot tears and shrinkage porosity.
During pouring, the liquid level in the pouring basin was maintained at a height of 3 to 5 times the sprue diameter. This constant head pressure prevents the formation of a vortex that can aspirate air and produce slag. The pour was stopped as soon as the risers were filled, and the pouring time was kept close to the calculated 160 seconds.
The pouring basin with a dam was used to hold back the slag. In addition, I used a ceramic foam filter with a pore size of 10 ppi directly below the sprue to further clean the melt. The combination of good pouring practice and effective filtration was key to achieving the 100% fluorescent MT requirement.
10. Numerical Simulation and Verification
Before the actual production trial, I ran a full 3D simulation using MAGMAsoft. The objective was to verify the filling pattern, temperature distribution, and porosity prediction. The mold filling simulation provided the velocity field at any instant. The maximum velocity in the ingates was less than 0.5 m/s, and the liquid metal front was smooth and uniform. The solidification simulation showed a temperature gradient directed toward the risers, with the last hot spots located inside the riser sleeves.
One important simulation result was the distribution of cooling rates in the heavy flange. The 200 mm thick section had a solidification time of approximately 120 minutes. In such a long solidification process, the risk of carbide formation and graphite degeneration is high. To mitigate this, I increased the inoculation level and added small chill blocks at selected locations to refine the microstructure.
The simulation also predicted a very low shrinkage porosity index in the critical areas. The predicted maximum porosity was less than 0.5%, which met the requirement for UT class 2.
11. Production Results and Inspection
After the simulation and process optimization, we conducted the actual production. The casting was poured, cooled, shaken out, and subjected to a full inspection program. The results are summarized below.
11.1 Dimensional Inspection
The casting was measured with a 3D laser tracker system. The results showed that all dimensions were within CT11 tolerance. Wall thickness measurements were within CT12. The machining allowance was sufficient, and no excess deviation was found. The weight of the as-cast part was within ±2.8% of the theoretical weight, satisfying the ±3% requirement.
11.2 Non-Destructive Testing
The casting underwent 100% ultrasonic testing (UT) according to EN 12680-3. The critical areas, including the rotor flange and the transition zone, were graded class 2, while non-critical areas were class 3. No recordable indications were found. The fluorescent magnetic particle testing (MT) also passed cleanly, with no surface defects exceeding the acceptance criteria.
11.3 Mechanical Properties
Table 4 shows the mechanical properties obtained from the attached test block.
| Property | Obtained Value |
|---|---|
| Tensile strength / MPa | 378 |
| Yield strength / MPa | 239 |
| Elongation / % | 20.5 |
| Hardness / HB | 143 |
| Impact energy at -20 °C, average / J | 14.0 |
| Impact energy at -20 °C, single values / J | 14.2, 14.0, 13.8 |
Table 4. Mechanical properties of the attached test block.
These values are well above the minimum requirements. The elongation of 20.5% indicates an extremely ductile ferritic matrix, which is essential for low-temperature toughness. The impact energy at -20 °C averaged 14 J, which is much higher than the 10 J minimum.
11.4 Metallurgical Examination
Metallographic samples were taken from both the attached test block and the casting body. The microstructures are shown in the attached photos. The graphite nodularity was evaluated according to ISO 945. The test block had a nodularity of >95% with graphite form V and VI dominating. The graphite size was class 5. The matrix was almost fully ferritic, with a few small pearlite areas less than 5% by volume. The casting body also met the requirements, with nodularity above 85% and pearlite content less than 12%.
The nodule count in the thick section (200 mm) was measured as approximately 70 nodules/mm², which is acceptable for such heavy sections. The absence of carbides and intercellular inclusions confirms the effectiveness of the melt treatment and inoculation.
12. Discussion
The successful production of this large main shaft demonstrates that with careful process design and rigorous control, it is possible to manufacture ductile iron castings that meet the most stringent low-temperature requirements. I would like to highlight several key lessons that are applicable to all producers of heavy-section ductile iron castings.
First, the gating system must be designed to avoid turbulent flow. A completely filled runner with a large unpressurized ratio and high filtration capacity is mandatory. The use of ceramic foam filters is not optional; it significantly improves the cleanliness of the casting and reduces the risk of dross defects.
Second, the feeding system must be designed with a full understanding of the solidification mechanism of ductile iron. Because of graphite expansion, it is not necessary to provide a riser that feeds all the solidification shrinkage, but one must ensure that the mold is rigid and that the riser is placed at the hottest point. The modulus ratio I used has been verified to produce sound castings in numerous applications.
Third, the chemical composition should be carefully optimized for low-temperature impact toughness. The combination of low manganese, low phosphorus, and controlled silicon is the basis of high toughness. In addition, the residual magnesium should be kept on the lower side to minimize dross formation, but still high enough to guarantee full nodularization.
Fourth, inoculation is an art. For large ductile iron castings, a single inoculation is rarely enough. The principal of “late inoculation” (mold stream or in-the-mold inoculation) is crucial to achieve a high nodule count and to prevent carbide formation in heavy sections. The use of a combination of silicon-based inoculants with small additions of bismuth and rare earths can greatly improve the nucleation potential.
Fifth, simulation is an indispensable tool. The numerical simulation allowed me to visualize the temperature field and flow pattern, and to identify the risk zones for shrinkage and oxidation. However, simulation results should always be validated by actual production data. In this case, the simulation predictions matched well with the UT results, which gives confidence in the reliability of the model.
Finally, the experience gained from this project confirms that ductile iron castings are capable of being used in very demanding structural applications, such as the main shaft of an offshore wind turbine. The combination of good design, high-quality melt, and proper casting practice yields products that are not only economical but also reliable under extreme conditions.
13. Conclusion
In this paper, I have presented a comprehensive overview of the casting production process for a low-temperature ductile iron main shaft used in an ultra-high-power wind turbine. The key features include a well-designed unpressurized gating system with effective slag skimming, a modulus-based riser system with insulated exothermic sleeves, and a strictly controlled melting and inoculation process. The use of numerical simulation significantly de-risked the process and allowed us to achieve a first-time-right result.
The final casting met all the requirements of the customer, including mechanical properties, metallographic structure, dimensional accuracy, and internal soundness. The impact toughness at -20 °C averaged 14 J, far exceeding the 10 J minimum. This success demonstrates the feasibility of manufacturing heavy-section low-temperature ductile iron castings with high quality.
I believe that ductile iron castings will continue to play a major role in the wind power industry, especially in the trend toward even larger turbines. As the size increases, the challenge of avoiding shrinkage and achieving toughness will become even greater. Future research should focus on improving the metallurgical quality in ultra-heavy sections, exploring new inoculants with lower fade rates, and advancing simulation models that better predict microstructural evolution. Nevertheless, the methodologies presented in this article provide a solid foundation for the production of reliable and high-performance ductile iron castings.
By sharing my practical knowledge and production data, I hope to contribute to the continuous improvement of foundry technology for large wind power components.