The production of critical, large-scale casting parts for the energy sector, such as the upper crown for hydraulic turbines, presents a unique set of formidable engineering challenges. As the demand for clean hydroelectric power grows, so does the physical scale of the turbines, pushing the dimensions and weights of their core casting parts beyond the limits of traditional foundry practices. The conventional approach of using a full, solid wooden pattern for two-part molding becomes increasingly untenable. This method, while straightforward for smaller components, fails catastrophically when applied to casting parts with diameters exceeding 7 meters and weights over 60 tonnes. The primary bottlenecks include the prohibitive cost and extended lead time for manufacturing the massive pattern itself, the extreme difficulty and risk in handling and flipping multi-hundred-ton sand molds, and significant logistical issues in pattern storage and transportation. This article details a first-person perspective on the development and implementation of an innovative, segmented three-part molding methodology that systematically deconstructs these challenges, enabling the feasible and cost-effective production of these monumental casting parts.
The cornerstone of our innovative approach is the strategic decomposition of the monolithic molding problem into three manageable segments: the lower mold (drag), the middle mold (check), and the upper mold (cope). Each segment employs a tailored formation technique—solid pattern molding, core assembly, and scraper molding, respectively—that plays to its geometric and logistical strengths. This segmentation directly attacks the core issue of mold weight and handling. For the lower section of the casting parts, which features a relatively smaller cross-section, we retain the use of a solid pattern. However, to circumvent the critical flip-weight limitation, we employ a nested flask strategy. The pattern is placed within a smaller, manageable flask (e.g., 4.5m x 4.5m) for initial molding. After this compact unit is rammed and successfully flipped—a safe operation due to its controlled weight—it is strategically placed and integrated into the final large-diameter production flask. Sand is then poured around it to complete the full lower mold footprint. This clever nesting technique bypasses the need to ever lift the entire mass of sand for the lower mold section in one piece.
The middle section of these large-diameter casting parts typically forms a conical shell. Creating this as a solid sand mold would be impractical. Instead, we fabricate it as an assembly of precisely manufactured sand cores. The number of segments is a calculated trade-off; too few and the cores become unwieldy, too many and assembly complexity increases. For a part with approximately 7.1m outer diameter, eight identical segment cores often provide the optimal balance. These cores are produced in reusable core boxes, ensuring dimensional accuracy and repeatability. They are then lowered into precise locators prepared in the lower mold and arranged in a circle. The middle flask is positioned around this assembly, and the remaining volume between the cores and the flask wall is filled with sand. This core-assembly method transforms an impossibly large, single-piece mold section into a kit of manageable, high-precision components.
The upper section, representing the large, sweeping inner surface of the crown, poses the greatest challenge for pattern-making. A solid pattern for this massive, complex curvature would be astronomically expensive and fragile. The traditional solution is rotational scraper molding, where a profile board is swept around a central axis to carve the shape from a sand bed. However, for casting parts of this scale, even this method has severe drawbacks: the scraper resistance is enormous, the supporting axis and beam framework become massive structures themselves, and the resulting single-use sand form is fragile. Our breakthrough was the development of a Segmented, Axis-Free Scraper Molding technique. Instead of one continuous sweep, we create a series of precision, reusable pattern segments that define the final surface. Key locators are placed on the middle mold. Removable segment patterns, representing slices of the final geometry, are then positioned between these locators. Sand is rammed against these segments to form robust, individual sand blocks. After curing, the segment patterns are removed, leaving behind a perfectly shaped, multi-piece sand mold that is both dimensionally accurate and structurally sound. This block assembly can be reused for multiple castings, a significant advantage over traditional one-time scraper molds.
| Methodological Aspect | Traditional Solid Pattern (Two-Part) | Innovative Segmented Three-Part |
|---|---|---|
| Pattern Cost & Lead Time | Extremely High (Full-size wood pattern) | Dramatically Reduced (Small pattern + core boxes + segment tools) |
| Mold Weight Handling | Impossible to flip/High collapse risk | Managed via nested flasks and segmentation |
| Operational Complexity | Simple in theory, impossible in practice for this scale | Higher planning, simpler execution per segment |
| Pattern/Mold Reusability | Solid pattern reusable, but storage is a major issue | Core boxes and segment tools highly reusable; sand molds more robust |
| Dimensional Control | Dependent on large, potentially flexible pattern | Enhanced via precise core boxes and segment tools |
| Suitability for Large Casting Parts | Not suitable for diameters >~5m | Specifically designed for diameters >5m up to 10m+ |
The selection of parting lines is a critical analytical step in this segmented methodology, directly influencing the weight distribution and feasibility of each mold section. Let the total height of the casting parts be $H_{total}$ and its maximum radius be $R_{max}$. The primary parting between the lower solid pattern section and the middle core assembly section occurs at a height $h_1$ from the bottom. The weight of the lower sand mold ($W_{drag}$) is a function of the sand density ($\rho_{sand}$) and the volume enclosed by the flask up to height $h_1$. By choosing $h_1$ such that the lower mold’s radius $R(h_1)$ is sufficiently small, we ensure $W_{drag}$ remains below the safe flipping capacity of our crane equipment, a constraint expressed as:
$$ W_{drag}(h_1) = \rho_{sand} \cdot V_{flask}(h_1) \leq W_{crane\_safe} $$
Conversely, the middle core assembly must form the remaining frustum-shaped volume. The complexity and cost of the core boxes increase with the size of the individual cores. Therefore, $h_1$ is not minimized arbitrarily but optimized to balance the lower mold weight against middle core manufacturability. A second key equation governs the subdivision of the middle conical section into $n$ identical segment cores. The arc length $s$ of each core at the largest diameter must be manageable for core shooting and handling:
$$ s = \frac{2 \pi R_{max}}{n} $$
An optimal $n$ is chosen so that $s$ corresponds to a core dimension that fits standard core-making equipment while minimizing the number of joints to be sealed during assembly. For a 7.1m diameter casting parts, $n=8$ gives $s \approx 2.79m$, a workable size.
| Parameter | Symbol | Example Value / Calculation | Rationale |
|---|---|---|---|
| Total Casting Weight | $W_{casting}$ | ~63,000 kg | Target component mass. |
| Lower Mold Parting Height | $h_1$ | ~1,200 mm | Chosen to keep drag flip weight ~75-80 tonnes. |
| Lower Pattern Diameter at $h_1$ | $D(h_1)$ | ~3,754 mm | Defines the size of the solid pattern required. |
| Number of Middle Section Cores | $n$ | 8 | Optimized for core box size and assembly. |
| Estimated Pattern Cost Reduction | – | >50% | Replacing full 7m pattern with 3.7m pattern + tools. |
| Estimated Pattern Lead Time Reduction | – | ~25 days | Avoiding construction of massive single pattern. |
The final assembly and solidification of these gigantic casting parts require meticulous control. Once the three sand mold sections are prepared, the internal core defining the hollow interior of the crown is lowered into the lower mold. All joints between mold sections and cores are carefully sealed with pastes or tapes to prevent metal penetration. The gating system, designed for balanced, progressive filling to minimize turbulence and thermal shock in the thin-walled structure, is integrated into the lower mold segments. Risers and feeders are strategically placed on the upper segmented mold to ensure soundness in the thickest sections of the casting parts. The pouring of several tens of tons of molten steel, typically a low-carbon martensitic stainless grade like ASTM A743 CA6NM, is a carefully choreographed operation. After cooling, the mold is broken away, revealing the raw casting. The segmented molding method inherently provides better dimensional accuracy due to the use of rigid, reusable tooling for cores and upper mold segments, reducing the variability associated with a single, large flexible wooden pattern.
In conclusion, the segmented three-part molding methodology, integrating nested solid pattern molding, precision core assembly, and innovative axis-free segmented scraper molding, represents a paradigm shift for manufacturing ultra-large, thin-walled rotational casting parts. It systematically solves the cardinal problems of weight, cost, and logistics that render traditional methods obsolete at this scale. The quantitative benefits are clear: drastic reductions in pattern cost and lead time, elimination of unsafe mold handling scenarios, and improved dimensional consistency. This approach not only makes existing designs producible but also expands the feasible design envelope for next-generation hydraulic turbines and other heavy machinery requiring massive, complex casting parts. The principles of decomposition, segment-specific optimal processes, and reusable tooling developed here provide a robust framework for tackling the ever-growing scale challenges in heavy industrial casting.