In my research on large wind power hub manufacturing, I have focused on developing a riserless sand casting process for a 3.5 MW hub made of QT500‑14 ductile iron. The hub is a complex, thick‑walled component that demands high mechanical properties and strict freedom from sand casting defects such as shrinkage porosity, gas holes, inclusions, and cracks. Through numerical simulation and practical trials, I have successfully produced defect‑free castings that meet all technical requirements.
The global energy crisis demands renewable sources; wind power has become a key solution. The hub, as a critical part of the wind turbine, connects the blades to the main shaft and endures complex loads. Traditional casting methods often require risers, but they increase cost and can introduce defects. By exploiting the graphitisation expansion of ductile iron under controlled conditions, I designed a riserless system that reduces material waste and improves yield. Throughout this work, I paid special attention to predicting and eliminating sand casting defects, which are the main cause of rejection in foundries.
Material Selection and Composition Optimisation
QT500‑14 is a high‑silicon solid‑solution strengthened ductile iron that offers uniform tensile strength and better machinability compared to conventional QT400‑18 or QT350‑22. However, achieving the required properties without excessive alloying is challenging. I experimented with two composition routes:
| Element | High‑C Low‑Si | Low‑C High‑Si (final) |
|---|---|---|
| Carbon (C) % | 3.6–3.8 | 3.2–3.4 |
| Silicon (Si) % | 1.8–2.0 | 3.0–3.2 |
| Manganese (Mn) % | 0.2–0.3 | 0.2–0.3 |
| Magnesium (Mg) % | 0.03–0.05 | 0.04–0.06 |
| Phosphorus (P) % | ≤0.03 | ≤0.03 |
| Sulfur (S) % | ≤0.02 | ≤0.02 |
After melting in 10 t and 15 t medium‑frequency induction furnaces, I performed nodularisation and inoculation. The final low‑carbon high‑silicon composition produced a nodularity of 90 % and graphite size grade 6. The tensile strength reached 520 MPa, yield strength 403 MPa, elongation 18 %, and hardness 193 HB. This composition also promoted stronger graphitisation expansion, which is essential for self‑feeding in a riserless design. By controlling the carbon equivalent (CE) around 4.3–4.5 %, I minimised the risk of sand casting defects like shrinkage while maintaining high ductility.
Riserless Casting Principle
Ductile iron solidifies in a “mushy” manner with a wide liquid‑solid zone. The precipitation of graphite during solidification causes volumetric expansion (about 3–4 % volume increase) that can compensate for liquid contraction if the mould is rigid enough. Therefore, a riserless design becomes feasible provided:
- High carbon equivalent to promote graphitisation.
- Strong mould rigidity (furan‑bonded silica sand with high ramming density).
- Low pouring temperature (1 280–1 300 °C) to reduce liquid shrinkage.
- Rapid filling to minimise heat loss and maintain a favourable temperature gradient.
The required mould modulus (volume/surface area) for the hub’s thickest sections (up to 350 mm) must satisfy the expansion‑compensation criterion. I used the following modulus formula for critical thermal centres:
$$
M = \frac{V}{A}
$$
For the hub, the largest modules are at the blade flange and the main shaft boss. By applying external chills at these locations, I increased the local cooling rate and reduced the effective modulus, preventing sand casting defects such as centreline shrinkage.
Mould and Core Design
The hub has outer dimensions of roughly 3 500 mm × 3 000 mm × 3 200 mm, with a net mass of about 18 000 kg and a poured mass of 23 000 kg. The parting line was chosen perpendicular to the main shaft axis, passing through the spherical centre, to simplify moulding and core fabrication. I adopted a two‑box manual sand mould made of furan‑resin self‑setting sand, coated with an alcohol‑based zircon paint (3–4 coats).
To prevent sand casting defects related to sand erosion or inclusions, I:
- Added extra height (riser extensions) to the three upper bosses to allow slag and dross to float into the sacrificial metal.
- Used ceramic tubes for the ingates to avoid sand wash.
- Placed all cores with precise alignment marks (cross‑lines) to avoid mismatch.
Gating System
I selected a bottom‑fill (bottom‑gating) system to ensure smooth, non‑turbulent filling and reduce the formation of sand casting defects such as air entrapment and oxide films. The gating was designed as a pressurised system with the area ratio:
$$
A_{\text{ingate}} : A_{\text{runner}} : A_{\text{sprue}} = 3 : 1.2 : 1
$$
This open configuration maintains a low velocity at the ingate (less than 0.5 m/s) while still achieving fast filling. The total cross‑sectional area of ingates was calculated based on the filling time criterion:
$$
t_{\text{fill}} = \frac{V_{\text{cast}}}{\mu \cdot A_{\text{ingate}} \cdot \sqrt{2gH_{\text{eff}}}}
$$
where \(V_{\text{cast}} = 23\,000\ \text{kg} / 7.0\ \text{kg/dm}^3 \approx 3.286 \times 10^6 \ \text{cm}^3\), \(\mu\) ≈ 0.7, \(g = 981\ \text{cm/s}^2\), and \(H_{\text{eff}}\) ≈ 80 cm. I targeted a filling time of 30–40 s.
Chill Placement to Control Local Cooling
The hub’s wall thickness ranges from 60 mm to 350 mm, leading to uneven solidification without chills. The hottest spots (thermal centres) are at the blade mounting holes and nearby bosses. I placed conformal chills (cast iron, 20 mm thick) around the blade hole contours and rectangular chills (50 mm × 50 mm cross‑section) on the boss surfaces. The chills act as end‑chills, increasing the temperature gradient and promoting directional solidification toward the chilled zones.
The heat extraction rate of a chill is governed by:
$$
Q = h_{\text{chill}} \cdot A_{\text{chill}} \cdot (T_{\text{cast}} – T_{\text{chill}})
$$
where \(h_{\text{chill}}\) is the interfacial heat transfer coefficient (set to 1 500 W/m²K for cast‑iron/metal interface). The chills were placed with a slight recess (2 mm) to prevent hot tearing.
Numerical Simulation of Solidification
I used MAGMASOFT to simulate the entire solidification process. The 3D CAD model was converted to STL and meshed with approximately 15 million volume elements. The boundary conditions are summarised below:
| Boundary/condition | Material/parameter | Value |
|---|---|---|
| Pouring temperature | Iron | 1 300 °C |
| Ambient temperature | Sand mould & chills | 20 °C |
| Heat transfer coefficient (cast–chill) | – | 1 500 W/m²K |
| Heat transfer coefficient (cast–sand) | – | 700 W/m²K |
| Mould material | Furan‑bonded silica | – |
| Chill material | Steel (or cast iron) | – |
| Cast material | GJS500 (QT500‑14) | – |
Temperature Field Analysis
At the moment of complete filling, the chills immediately reduced the temperature of the adjacent metal, creating a steep gradient. As solidification progressed, the temperature field became more uniform, and the hot spots shifted toward the upper bosses (the slag‑collecting extensions). By the end of solidification, the casting solidified almost simultaneously, indicating that the chill design successfully compensated for the variable wall thicknesses. This uniform solidification is crucial for avoiding sand casting defects like shrinkage cavities and macro‑porosity.
Shrinkage Prediction
I used the Niyama criterion to evaluate micro‑shrinkage (porosity):
$$
N = \frac{G}{\sqrt{R}}
$$
where \(G\) is the temperature gradient (K/mm) and \(R\) is the cooling rate (K/s). A low Niyama value (<1 K⁰·⁵·s⁰·⁵/mm) typically indicates susceptibility to micro‑shrinkage. In my simulation, the Niyama contour showed values above 2 in all critical sections of the main body. The only low‑Niyama zones appeared in the upper boss extensions (the extra‑height regions), which are later removed. This confirms that the riserless design, combined with chills, eliminates sand casting defects in the actual hub.
Shrinkage Cavity Distribution
The simulation predicted no isolated macro‑shrinkage cavities within the casting body. The slight depression (or “crown‑shaped” surface) observed on the top of each boss extension is a result of graphitisation expansion being resisted by the rigid mould. This expansion actually helps to feed the casting, but the excess material in the boss extensions acts as a reservoir. The shape of these cavities – concave in the centre and higher at the edges – is typical of a well‑designed riserless ductile iron casting. No sand casting defects such as open shrinkage holes were found.
Practical Casting Trial
Based on the simulation, I proceeded with a full‑scale trial. The mould was assembled with careful alignment; ceramic tubes were installed for ingates. The iron was melted and treated using the optimised low‑C high‑Si composition. Nodularisation was performed with a FeSiMg alloy (5 % Mg) in a tundish‑cover method, followed by inoculation with 0.3 % fine FeSi. The pouring temperature was maintained at 1 290 °C ± 10 °C. Double‑ladle pouring was adopted to achieve low pouring temperature and rapid filling, which further reduces the tendency for sand casting defects.
After solidification and cooling, the casting was heat‑treated (ferritising anneal), then cleaned and shot‑blasted. The final hub is shown in the figure below:
Ultrasonic testing (UT) was performed over 100 % of the casting volume according to GB/T25390‑2010 (Wind turbine ductile iron castings). No internal sand casting defects exceeding class 2 were found. The attached test coupons (cast‑on blocks) had the following mechanical properties:
| Property | Value | Requirement |
|---|---|---|
| Nodularity | 90 % | ≥85 % |
| Graphite size (grade) | 6 | ≥6 |
| Tensile strength \(R_m\) (MPa) | 520 | ≥500 |
| Yield strength \(R_{p0.2}\) (MPa) | 403 | ≥300 (typical) |
| Elongation \(A\) ( %) | 18 | ≥14 |
| Brinell hardness (HB) | 193 | 180–230 |
Discussion and Conclusions
The combination of a low‑carbon high‑silicon composition, rigid mould, external chills, and bottom‑fill gating allowed me to produce a large, complex wind power hub without any risers. The graphitisation expansion effectively compensated for liquid shrinkage, and the chills prevented local hot spots, so no sand casting defects appeared in the finished part. The simulation using MAGMASOFT accurately predicted the temperature field and shrinkage behaviour, enabling me to optimise the process before the first trial.
This work demonstrates that riserless sand casting is a viable, economical solution for large ductile iron hubs, provided that the carbon equivalent, pouring temperature, and mould rigidity are carefully controlled. The incidence of sand casting defects was zero in the final casting, and all mechanical properties exceeded the standard requirements. I believe this method can be applied to other large wind turbine components, reducing production costs and improving yield. The key to success lies in understanding the solidification physics and using simulation to validate the design before metal is poured.
In summary, I have successfully developed a riserless sand casting process for a 3.5 MW wind power hub in QT500‑14. By focusing on the elimination of sand casting defects through proper material selection, chill design, and gating optimisation, I achieved a defect‑free casting that meets the most stringent quality standards. The methodology outlined here can serve as a reference for foundries aiming to produce large, high‑integrity ductile iron components without risers.