As a researcher focused on advanced casting techniques, I have extensively studied the production of large, complex stainless steel impellers used in critical applications such as pharmaceutical equipment. These impellers operate in highly corrosive environments and demand exceptional precision, surface finish, and mechanical integrity. The impeller in question features a diameter of 700 mm, a height of 130 mm, a net weight of 202 kg, and 54 intricately twisted blades with a thickness of only 5 mm. Traditional methods like sole investment casting or sand casting alone proved insufficient due to challenges in achieving both dimensional accuracy for the blades and effective feeding for the heavy sections like the hub and spokes. This led me to develop a hybrid approach: a compound process combining investment casting for the blade region and sand casting for the simpler hub and spoke sections—the core sand casting parts. This methodology leverages the high precision of investment casting for complex geometries and the cost-effectiveness and feeding capability of sand casting for bulkier components.
The fundamental challenge was to integrate these two distinct casting methods seamlessly. The blade assembly, with its tight spacing and aerodynamic profiles, required a precise mold that could only be economically achieved through investment casting. Meanwhile, the hub and spokes, being relatively simple shapes but containing significant thermal masses, were ideal for sand casting, which allows for robust gating and risering systems. The key innovation was dividing the impeller at an inner diameter of 416 mm, creating a clear interface. The investment-cast shell would form the blades and outer rim, while the sand mold would form the central hub, spokes, and the backside of the blades, effectively encapsulating the investment shell. This compound mold strategy resolved issues of mold-making complexity and solidification feeding simultaneously.
In designing the overall casting plan, I adopted a horizontal parting line with a two-flask system to simplify pattern making and molding operations. The gating and risering design was critical due to the high volumetric shrinkage of the ZG1Cr18Ni9Ti austenitic stainless steel material. A top-pouring system was chosen to ensure rapid filling and maintain a favorable temperature gradient for feeding. The central hub was fed by a single large riser that also served as the main pouring cup, while three additional risers were placed equidistantly around the outer rim to feed the blade root junctions. The modulus method was employed for riser sizing. For the hub riser, the modulus \( M_{hub\_riser} \) was calculated based on the geometry of the hub junction. The modulus \( M \) of a casting section is given by \( M = V/A \), where \( V \) is volume and \( A \) is cooling surface area. For a cylindrical hub section, approximations yield a modulus. After calculation, a standard elliptical riser with dimensions 120 mm width, 180 mm length, and 150 mm height was selected. Its feeding capacity \( V_C \) was verified using the formula for volumetric shrinkage \( \epsilon \):
$$ V_C = \frac{V_{riser} \cdot (1 – \epsilon)}{\epsilon} $$
where \( \epsilon = 6\% \) for this steel. Similarly, for the rim risers, a modulus \( M_{rim\_riser} = 2.27 \, cm \) led to the selection of risers with dimensions 110 mm width, 165 mm length, and 150 mm height. Feeding distance checks confirmed that three such risers were necessary. The following table summarizes the key riser parameters:
| Riser Location | Modulus (cm) | Dimensions (W×L×H, mm) | Calculated Feeding Volume \( V_C \) (cm³) | Number Used |
|---|---|---|---|---|
| Hub (Central) | 2.48 | 120×180×150 | 4.3×10³ | 1 |
| Rim (Peripheral) | 2.27 | 110×165×150 | 3.0×10³ | 3 |
The machining allowances were set at 5 mm for the top face and 3 mm for the bottom face, accounting for any misalignment between the investment shell and the sand mold. For the sand casting parts, particularly the core forming the inner passages, a 5 mm allowance per side was allocated.
The creation of the investment casting shell for the blades was a meticulous process. A low-shrinkage rosin-wax pattern material with a contraction rate of 0.75% was used. The pattern die was machined from 45# steel with a cavity surface roughness of Ra 0.8–0.2, and a total pattern contraction allowance of 2.6% was incorporated. To assemble the 54 individual blade patterns accurately into a complete ring, I designed a specialized welding fixture. This fixture consisted of top and bottom plates, a precision locating ring, and a central clamping shaft. Each blade pattern, with its specific root contour, was positioned against the locating ring, and the assembly was verified using templates to ensure chord length differences between adjacent blades were within 3 mm and twist angle deviations were under 1°. This fixture was also crucial during the subsequent shell-building process, as it provided a stable platform for coating.
Given the large size and complex geometry of the blade cluster, a hybrid ceramic shell system was developed to balance strength, cost, and surface finish. A silica sol-based primary coat was applied for high refractoriness and detail reproduction, followed by intermediate and backup coats using a modified sodium silicate binder system for faster build-up and lower cost. The coating process involved successive dips in ceramic slurries and stuccoing with refractory sands. After the required coat thickness was achieved, the assembly, still in its fixture, underwent a reinforcement step. A 100 mm thick shell of CO2-hardened sodium silicate sand was built around the peripheral investment shell. Within this sand reinforcement shell, pre-cast iron braces with lifting lugs were embedded. This served dual purposes: it prevented distortion of the delicate ceramic shell during handling and subsequent dewaxing, and it provided secure lifting points and alignment features for later mold assembly. The reinforced shell was then dewaxed using high-pressure steam at 0.6–1.0 MPa for 6–10 minutes. Finally, firing was conducted at 850°C for 1.5–2 hours in an electric furnace, with the shell placed horizontally in a tray of supporting sand to prevent sagging.
Concurrently, the sand casting parts—the mold for the hub, spokes, and the cavity that would accept the investment shell—were produced. A one-piece wooden pattern with a 2.2% shrinkage allowance was used for the drag (lower) mold. The pattern included the impression for the hub, spokes, and a recess matching the outer contour of the fired investment shell. For the core forming the internal features, a split wooden core box was employed. The molding material selection was critical for the sand casting parts facing the high-temperature steel. The facing sand was 100% chromite sand of 40/70 mesh, mixed with sodium silicate binder and hardened by CO2 gassing. Chromite sand offers excellent chilling power and resistance to metal penetration, which is vital for the thin blade sections cast against it. The backing sand was conventional silica sand. The mixed sand was compacted around the pattern, and ventilation holes were prudently poked in the cope section above each blade location to facilitate gas escape during pouring. Gassing parameters were typically a pressure of 0.10–0.15 MPa for 20–30 seconds. After stripping, the mold surfaces in contact with the metal were coated with a zircon-based wash to further enhance surface finish.
The final and most critical step was the assembly of the compound mold. The fired and reinforced investment shell, now representing the exact negative of the 54 blades and outer rim, was carefully lowered into the precisely machined recess in the drag sand mold. Alignment was ensured by the combination of the locating features on the iron braces and guide pins in the flask. The cope (upper) sand mold, containing the impressions for the risers and the top surfaces of the hub and spokes, was then placed on top. The interface between the ceramic shell and the sand molding material for the sand casting parts had to be perfectly sealed to prevent metal penetration. The completed mold assembly created a unified cavity: the intricate blade passages were defined by the ceramic, while the volumous hub, spokes, and the back of the blades were defined by the sand. The pouring basin was integrated into the central hub riser in the cope.
The melting of ZG1Cr18Ni9Ti was carried out in a medium-frequency induction furnace with a basic lining. The charge consisted of 60% returns, 20% low-phosphorus steel scrap, and necessary ferroalloys (chromium, nickel, manganese, silicon, and titanium). Deoxidation was performed using aluminum additions. The target pouring temperature was 1580 ± 10°C. Pouring was rapid initially to ensure complete filling of the thin blade sections, then slowed towards the end, with a total pouring time of 8–12 seconds. The risers were kept full to promote directional solidification from the blade tips towards the hub and rim risers.
After cooling, the casting was knocked out. The sodium silicate sand shell for the sand casting parts was easily removed, but the ceramic shell around the blades required chemical cleaning via an alkaline boil-out process to avoid mechanical damage to the delicate blade surfaces. The large risers, being part of the sand casting parts, were removed using oxyacetylene flame cutting with vibration to overcome the work-hardening tendency of the austenitic stainless steel. Heat treatment was essential to achieve optimal corrosion resistance and mechanical properties. A solution treatment at 1050–1100°C for 3–4 hours followed by water quenching was conducted to dissolve carbides and homogenize the structure. This was followed by a stabilization treatment at 850–950°C for 4 hours and air cooling to prevent sensitization and enhance dimensional stability.
Quality assessment of the produced impellers confirmed the success of the compound process. Dimensional inspection showed overall accuracy levels conforming to CT4–CT5, with blade profile tolerances (chord length differences and twist angles) well within specified limits. Surface roughness on the blades measured Ra 3.2–1.6 µm. Chemical analysis confirmed the material met ZG1Cr18Ni9Ti specifications. Mechanical testing revealed yield and tensile strengths exceeding standard requirements, and the castings were non-magnetic after heat treatment. Most importantly, all impellers passed stringent overspeed tests (110% rated speed for 2 minutes) and dynamic balancing trials. The process yielded a 100% success rate over multiple production runs.
The economic and technical advantages of this compound approach are significant. It dramatically reduces the cost compared to full investment casting of such a large part, as the volumous sand casting parts are produced by a much cheaper method. It also solves the feeding problem inherent in thin-walled, large-diameter investment castings by incorporating substantial sand-cast risers. The process stability is high because each sub-process—investment shelling for complexity and sand molding for bulk—operates within its optimal window. The integration of sand casting parts into a precision casting workflow opens new possibilities for manufacturing large, hybrid-geometry components. The following formula encapsulates the overall linear shrinkage consideration \( S_{total} \) for pattern making, which had to account for both investment and sand casting contributions:
$$ S_{total} = S_{material} + S_{molding\_process} $$
where \( S_{material} \) is the alloy shrinkage (approx. 2.2% for this steel) and \( S_{molding\_process} \) is an adjustment factor for process-specific constraints, which was empirically determined to be 0.4% for this compound setup, leading to the applied 2.6% pattern allowance.
Further analysis of the solidification dynamics can be modeled using Chvorinov’s rule. The solidification time \( t \) for a section is proportional to the square of its modulus:
$$ t = k \cdot M^2 $$
where \( k \) is the mold constant. In our compound mold, the thin blades (\( M_{blade} \approx 0.25 \, cm \)) solidified very quickly, while the hub junction (\( M_{hub} \approx 2.5 \, cm \)) solidified much slower. The designed risers with \( M_{riser} > M_{casting} \) ensured they remained liquid longest, creating the necessary feeding pressure. The table below contrasts the moduli of different sections, highlighting the need for the compound approach:
| Casting Section | Approximate Modulus, \( M \) (cm) | Primary Manufacturing Method | Key Challenge |
|---|---|---|---|
| Blade (airfoil section) | 0.25 | Investment Casting | Precision, surface finish |
| Blade root at rim | 1.8 | Investment Casting (interface with sand) | Hot spot, feeding |
| Hub and Spoke Junction | 2.5 | Sand Casting Parts | Volumetric shrinkage, feeding |
| Outer Rim (solid section) | 2.0 | Investment Casting (fed by sand risers) | Interface integrity |
In conclusion, the fusion of investment casting and sand casting technologies into a single compound process represents a powerful solution for large, complex components like stainless steel impellers. The method capitalizes on the precision of investment casting for intricate features and the robust feeding and economic benefits of sand casting for the massive sand casting parts. The development of specialized fixtures, hybrid shell systems, and integrated mold assembly protocols ensured dimensional fidelity and metallurgical soundness. This approach has broad applicability beyond impellers, potentially revolutionizing the production of other large components with detailed features, such as turbine hubs, pump casings, or architectural elements, where performance and cost must be optimally balanced. The successful implementation demonstrates that hybrid manufacturing strategies can overcome limitations inherent in single-process casting, paving the way for more innovative and efficient foundry practices.