In the early stage of new casting development, the required number of prototypes is usually small. Traditional sand casting methods necessitate investing in metal patterns, which are expensive and require a lead time of 1.5 to 3 months for manufacturing. During product development, design changes from customers often force modifications to the casting process or even the patterns themselves. When pattern alterations become difficult or impossible, they may be scrapped, resulting in significant economic loss. Moreover, the long cycle from process design to pattern completion means slow response to customer demands. In conventional sand casting, complex geometries often cannot be withdrawn from the mold, requiring numerous cores or loose pieces to reproduce the shape. This involves multiple core boxes, core assembly, and intricate operations, leading to low dimensional accuracy. 3D printing technology effectively overcomes these issues by enabling patternless casting and rapid sand mold fabrication. As an emerging technique, 3D printing offers clear advantages, being almost unrestricted by spatial constraints and capable of producing intricately shaped components. In my work, I combined 3D printed sand molds with simulation technology to achieve the rapid development of a multi-curved impeller body casting prototype.
Casting Structure Analysis
The impeller body casting is required by the customer to be made of QT500-7 ductile iron. Its maximum external dimensions are Φ660 mm × 202 mm, with a casting mass of 105 kg and a blade wall thickness of 21 mm. The center of the casting contains a large thermal center, and the six blades are twisted curved surfaces. Using the traditional assembled core method would require at least two sand cores per blade, totaling 12 cores or more. After assembly, the joint lines would produce extensive flash, making grinding difficult. Over-grinding could damage the smooth transition of the blade curved surfaces, compromising the performance of the impeller component. Furthermore, the large wall thickness variations inevitably lead to shrinkage defects, demanding a fast pouring process.
The following table summarizes the key geometrical and material parameters of the impeller body casting:
| Parameter | Value |
|---|---|
| Material | QT500-7 (Ductile Iron) |
| Maximum outer diameter | Φ660 mm |
| Height | 202 mm |
| Mass | 105 kg |
| Blade thickness | 21 mm |
| Number of blades | 6 (twisted curved) |
| Center cylinder thickness | 180 mm |
Gating and Riser Design
Considering the structural features and material properties of the impeller body, both rapid pouring and effective feeding of the heavy thermal center were essential. Three feeding schemes were evaluated through simulation:
- Single top riser: A single top riser positioned directly above the thick center cylinder (180 mm thickness). Simulation showed that a very large riser with a neck diameter almost equal to the cylinder diameter was required to draw shrinkage into the riser. Such a massive riser created an even larger artificial hot spot, leading to local overheating, coarse microstructure at the riser root, and difficult removal.
- Multiple ingates on blades: Placing ingates on each blade would disrupt the blade geometry, cause severe turbulence during filling, and create cleaning problems due to grinding marks on the blades.
- Two top risers with chills: Two small top risers were designed on the top surface of the center cylinder, but not at the exact center, to minimize artificial hot spots and facilitate removal. A chill was placed at the bottom of the center cylinder to promote directional solidification. Foam ceramic filters were placed on top of both risers to filter the melt, reduce flow velocity, and stabilize pouring. To avoid filter erosion, pouring was alternated between the two risers.
The final adopted scheme is illustrated conceptually: two small risers on the top and a chill at the bottom. The solidification simulation of this scheme predicted that shrinkage porosity and cavities would transfer into the risers. Although some minor shrinkage appeared at the riser neck root, experience indicated that because the risers are directly poured and thus overheated, these defects would be fully drawn into the riser body.
The theoretical feeding volume can be estimated using the modulus concept. For a riser to be effective, its modulus $M_R$ must be greater than the modulus of the casting region it feeds $M_C$. The modulus is defined as:
$$ M = \frac{V}{A} $$
where $V$ is the volume and $A$ is the cooling surface area. For the center cylinder, the modulus is large, requiring a riser with a suitably high modulus. The feeding efficiency $\eta$ for a ductile iron riser typically ranges from 14% to 20%. The required riser volume $V_R$ is:
$$ V_R = \frac{V_{shrink}}{\eta} $$
where $V_{shrink}$ is the volumetric shrinkage of the casting segment being fed. For ductile iron, the solidification shrinkage is about 4% to 5%. Using these formulas, the riser dimensions were optimized to ensure soundness.
| Scheme | Riser Type | Advantages | Disadvantages |
|---|---|---|---|
| 1 | Single large top riser | Simple design | Large artificial hot spot, coarse structure, difficult removal |
| 2 | Ingates on each blade | Distributed filling | Blade distortion, turbulence, cleaning problems |
| 3 | Two small top risers + bottom chill | Reduced artificial hot spot, effective feeding, easy removal | Requires alternating pouring |
The solidification simulation confirmed that the two-riser scheme with a chill successfully eliminated internal shrinkage defects. The simulated temperature distribution and porosity risk indicated that all shrinkage was confined to the risers, validating the design.
3D Printed Sand Mold Design
Based on the finalized casting process, the sand mold was designed for 3D printing. The parting line was set at the mid-height of the blades. The mold wall thickness (sand allowance) was 40 to 50 mm. Three sand cones were incorporated on the parting plane for accurate alignment. To reduce printing volume and material consumption, the mold wall thickness was minimized while maintaining sufficient strength. Handling and assembly recesses were added on the sides. Venting holes, not communicating with the mold cavity, were placed on the parting plane to allow gas escape during pouring. The casting shrinkage rate was set to 0.8%, and a coating thickness of 0.2 mm was pre-compensated in the digital model.
Using 3D printing, the entire sand mold for the impeller body required only two halves (upper and lower). In contrast, traditional sand casting would need more than 12 separate cores to form the twisted blades because the complex curvature cannot be withdrawn from a rigid pattern. The fewer the mold segments, the higher the dimensional accuracy, the less flash at joints, and the cleaner the casting surface. 3D printing enables patternless casting and flexible production, drastically reducing sample manufacturing cost and cycle time to about 20 days.
The following table highlights the benefits of 3D printed sand molds versus conventional core assembly for this impeller body:
| Item | Conventional (Core Assembly) | 3D Printing |
|---|---|---|
| Number of mold/core pieces | 12+ cores | 2 mold halves |
| Pattern/mold cost | High (metal pattern + core boxes) | Low (no pattern needed) |
| Lead time | 1.5–3 months | ~20 days |
| Modification flexibility | Difficult, expensive | Easy (modify digital model) |
| Dimensional accuracy | CT9–CT10 (cumulative errors) | CT7–CT8 |
| Flash at joints | Significant | Minimal |
Rapid Mold Fabrication
The impeller body casting was intended as a prototype for customer product development, with three pieces required. The digital sand mold model was sent to a 3D printing service provider for customized production. The sand cores were printed to a dimensional tolerance grade of CT7 to CT8 (ISO 8062). The required sand strength was a room-temperature tensile strength of at least 1.5 MPa. After receiving the printed sand molds, an alcohol-based zirconia coating was applied to the cavity surfaces with a thickness controlled between 0.15 and 0.20 mm, then ignited and dried. The chill was made of 45 steel; its working surface was also coated with the same alcohol-based coating and flame-heated before assembly.
An image of the 3D printed sand mold (before coating) is shown below. The intricate blade geometry is clearly visible, demonstrating the capability of additive manufacturing.
The application of coating and placement of the chill and foam ceramic filters were performed manually. The filters were placed on top of the two risers to reduce turbulence and entrainment, as shown in the process description.
Pouring and Machining Verification
To prevent run-out during pouring, the assembled sand mold was embedded in a flask with ballast sand. The total pouring weight was 126 kg. After pouring and solidification, the castings were knocked out, degated, ground, and shot-blasted. All three castings passed dimensional inspection, exhibited a clean surface, and met ultrasonic flaw detection requirements. The dimensional deviations were within the allowed tolerance of CT8. The machined samples were delivered to the customer and successfully passed the machining and functional tests.
The following table summarizes the quality inspection results for the three castings:
| Casting No. | Dimensional Accuracy (CT grade) | Surface Condition | Ultrasonic Test | Machining Verification |
|---|---|---|---|---|
| 1 | CT7–CT8 | Clean, minimal flash | No defects | Pass |
| 2 | CT7–CT8 | Clean | No defects | Pass |
| 3 | CT7–CT8 | Clean | No defects | Pass |
The similarity between the simulated and actual solidification behavior validated the process design. The finished impeller body samples after machining were fully functional and met customer satisfaction.
Conclusion
By integrating 3D printed sand molds with simulation technology, I successfully achieved the rapid development of a complex impeller body casting. Three consecutive castings were produced with excellent surface quality, dimensional accuracy, and internal soundness. The entire cycle from design to delivery was reduced to approximately three weeks, without any investment in metal patterns. This approach is particularly advantageous for sample and prototype production, enabling low-cost, high-precision, and fast manufacturing of geometrically complex castings.
3D printing for sand casting is a game-changer in the foundry industry. It eliminates pattern costs, shortens lead times, and offers unparalleled design freedom. For single-piece or small-batch production, especially when design iterations are anticipated, the combination of 3D printing and simulation provides an agile solution that traditional sand casting cannot match. The successful case of the impeller body demonstrates the broad application prospects of 3D printed sand molds for complex castings in the future.