In my research on the casting of a blower impeller, I focused on replacing the traditional wooden pattern with a pattern manufactured by fused deposition modeling (FDM) 3D printing. The traditional wooden pattern suffered from complex machining, low efficiency, high cost, and susceptibility to damage and cracking during use. By utilizing NX10.0 software for three-dimensional digital modeling and an UP box 3D printer, I printed the impeller pattern using PLA filament. This pattern was then employed in sand casting to produce the final aluminum impeller. The results demonstrated significant improvements in material utilization, manufacturing cycle time, and cost reduction, while maintaining satisfactory dimensional accuracy and surface quality.
Additive manufacturing, commonly known as 3D printing, builds physical objects layer by layer from a digital model. In sand casting, the pattern is a critical element that defines the cavity shape. Traditional wooden patterns require skilled carpentry and often involve complex machining steps such as turning, milling, and drilling, especially for parts with curved surfaces and thin sections like the impeller blades. 3D printing eliminates many of these steps by directly creating the pattern from a CAD file. The principle of FDM 3D printing is illustrated in the following figure, which shows the typical workflow from 3D modeling to solid model.
The impeller I studied is a component of a blower. Its geometry includes thin curved blades with a draft angle of 2° and a gradual thickness variation. The overall dimensions and features are defined in the engineering drawing. I created a three-dimensional digital model using NX10.0 software. The pattern was designed as a whole-mold with the parting plane located at the largest cross-section on the back side of the impeller. The casting material was aluminum alloy ZL301, which has a linear shrinkage of approximately 1.3%. To compensate for this shrinkage, I scaled the model accordingly. The blades were given a 2° draft to facilitate pattern removal from the sand mold. A top-gating system was chosen because the impeller has a relatively simple shape and thin walls; this type of pouring system helps ensure complete filling of the thin sections.
The manufacturing of the pattern via 3D printing began with exporting the solid model in STL format. I used Cura software to slice the model and defined two printing strategies for comparison. The printer used was an UP box FDM machine with a nozzle diameter of 0.4 mm. The filament material was polylactic acid (PLA) with a diameter of 1.75 mm. The two strategies differed in layer height, print speed, nozzle temperature, support type, and platform adhesion. Table 1 summarizes the parameters for each strategy.
| Parameter | Strategy 1 | Strategy 2 |
|---|---|---|
| Layer height (mm) | 0.2 | 0.1 |
| Wall thickness (mm) | 0.8 | 0.8 |
| Print speed (mm/s) | 50 | 40 |
| Nozzle temperature (°C) | 210 | 210 |
| Support type | Everywhere | Everywhere |
| Platform adhesion type | Raft | Brim |
| Printing time | 3 h 58 min | 6 h 59 min |
| Filament length used (m) | 12.19 | 10.59 |
| Filament weight (g) | 36 | 32 |
| Surface quality | Coarser on back side | Finer on back side |
After printing, I evaluated the surface roughness of the blades and the back side. Strategy 1 produced a slightly coarser texture on the back side compared to Strategy 2, but the blade surfaces were similar. Because the impeller operates inside a blower housing where surface finish is not critical, I selected Strategy 1 for its shorter printing time and acceptable quality. The total manufacturing time for the pattern, including 1 hour of 3D modeling and 4 hours of printing, was only 5 hours. In contrast, a traditional wooden pattern would require 1 hour for modeling, 1 hour for CNC programming, 1 day for fixture design and fabrication, and 6 hours of machining, totaling 32 hours.
The next step was sand mold preparation. I placed the printed PLA pattern with the back side down into the drag flask. Silica sand mixed with a clay binder was compacted around the pattern. After leveling the sand, I turned the flask over, dusted the parting surface with parting powder, and placed the cope flask on top. A sprue pin was positioned 30–40 mm away from the pattern. After compacting the cope sand, I removed the pin and created vent holes with a vent rod. I then opened the cope, brushed away any loose sand, and carefully extracted the pattern from both halves. Runners and gates were cut into the cope and drag to form the top-gating system. The mold was closed and clamped for pouring. The aluminum alloy ZL301 was melted at a pouring temperature of 755°C and poured into the mold. After solidification and shakeout, the casting was separated from the gating system. Figure 2 shows the final impeller casting after cleaning.
I inspected the casting dimensions and surface quality. All critical dimensions met the specifications, and the surface roughness was adequate for the intended application. The impeller was found to be suitable for blower service. To quantify the advantages of the 3D-printed pattern over the traditional wooden pattern, I compared material utilization, manufacturing cycle time, and processing cost. Table 2 presents the comparative data.
| Metric | 3D-printed pattern | Traditional wooden pattern |
|---|---|---|
| Material utilization | 98% (only support waste) | 46% (most material removed by machining) |
| Manufacturing cycle | 5 h (modeling 1 h + printing 4 h) | 32 h (modeling 1 h + programming 1 h + fixture design 1 day + machining 6 h) |
| Processing cost | 54 CNY (PLA filament 3.6 CNY + processing fee 50 CNY) | 1,600 CNY (wood 100 CNY + fixture 500 CNY + machining 1,000 CNY) |
| Durability | High hardness, low wear, resistant to moisture | Low hardness, easy to wear, prone to cracking and deformation due to moisture |
The data clearly show that the 3D-printed pattern increased material utilization from 46% to 98%, reduced the manufacturing cycle by a factor of 6.4, and lowered the cost by a factor of 29.6. Furthermore, the PLA pattern exhibited better mechanical properties and resistance to environmental conditions compared to wood. Wood patterns tend to absorb moisture, swell, and crack, while PLA remains stable. This stability extends the service life of the pattern and improves the consistency of the castings.
I also performed a theoretical analysis of the shrinkage compensation. The linear shrinkage of ZL301 aluminum alloy is 1.3%. The pattern dimensions were scaled using the formula:
$$ L_{pattern} = L_{casting} \times (1 + \frac{\epsilon}{100}) $$
where \(\epsilon = 1.3\%\). For example, if a casting dimension is 100 mm, the pattern dimension becomes 101.3 mm. The draft angle of 2° was applied to all blade surfaces to ensure easy withdrawal from the sand. The draft can be expressed as:
$$ \tan \alpha = \frac{\Delta t}{h} $$
where \(\alpha = 2^\circ\), \(h\) is the blade height, and \(\Delta t\) is the thickness reduction. For a blade height of 25 mm, the thickness reduction per side is approximately:
$$ \Delta t = h \cdot \tan 2^\circ \approx 25 \times 0.0349 = 0.87 \; \text{mm} $$
This ensures that the blade can be removed from the sand without damaging the mold cavity. The top-gating system was designed with a sprue height of 100 mm, a runner cross-section of 15 mm × 10 mm, and an ingate area of 20 mm × 3 mm. The filling time can be estimated using Bernoulli’s equation, but due to the complexity of the flow in thin sections, practical experience guided the design.
The mechanical properties of the PLA material used for the pattern were sufficient for sand casting. PLA has a tensile strength of about 50–60 MPa, a modulus of elasticity of 3.5 GPa, and an elongation at break of 2–5%. These properties are similar to or better than common hardwoods used for patterns. Additionally, PLA has a low thermal expansion coefficient (about 0.00007 /°C) compared to wood (0.00003–0.00006 /°C in the radial direction, but highly anisotropic). This dimensional stability is critical for maintaining pattern accuracy over repeated uses. The glass transition temperature of PLA is around 60°C, which is below the mold temperature during sand compaction (typically room temperature). Therefore, the pattern does not soften during the molding process.
Another important aspect is the surface finish of the pattern. The FDM process leaves visible layer lines. For a layer height of 0.2 mm, the theoretical surface roughness (Ra) is approximately 0.2 mm peak-to-valley. In practice, the actual Ra is around 0.1–0.15 mm after considering material flow and shrinkage. This roughness is adequate for sand casting, where the pattern surface is transferred to the sand mold and subsequently to the casting. Post-processing techniques such as sanding or vapor smoothing could further improve the finish if required, but for the impeller application it was not necessary.
I also considered the environmental and economic impact. 3D printing produces very little waste compared to machining. For the wooden pattern, the material utilization was only 46%, meaning more than half of the wood was turned into chips. In contrast, the 3D-printed pattern consumed only 36 grams of PLA filament, with the only waste being a small amount of support material. The energy consumption for printing was about 0.5 kWh, whereas machining a wooden pattern would require several kilowatt-hours for the CNC machine and additional energy for fixture fabrication. Thus, the 3D printing approach aligns with sustainable manufacturing principles.
Furthermore, the digital nature of the 3D printing process allows easy modification of the pattern. If the impeller design needs to be changed, I only need to update the CAD model and reprint. This flexibility is especially valuable for prototyping and small batch production where pattern modifications are frequent.
In addition to the FDM technology used in this study, other 3D printing processes such as selective laser sintering (SLS) of metal powders or binder jetting (3DP) of sand are also used in the foundry industry. SLS can produce high-strength metal patterns directly, but the cost of metal powder is high. Binder jetting of sand directly creates the mold without a pattern, but the equipment is expensive and maintenance-intensive. FDM offers an excellent balance of cost, speed, and ease of maintenance. The choice of process depends on the specific requirements of the casting, the batch size, and the budget.
I also explored the possibility of directly 3D printing the sand mold using the binder jetting technique. In that approach, a layer of sand is spread, and a binder is selectively deposited according to the cross-section of the part. After the entire mold is built, the unbound sand is removed, leaving the cavity. This method eliminates the need for a pattern altogether. However, for the impeller studied, the pattern method was chosen because the FDM printer was readily available and the sand mold could be prepared using conventional foundry equipment. The pattern also provides the advantage of being reusable for multiple castings, whereas a 3D-printed sand mold is used only once.
To further quantify the savings, I calculated the return on investment. The initial investment for an FDM 3D printer (such as the UP box) is approximately 8,000 CNY. The cost per print (including filament and electricity) is about 5 CNY per hour of printing. For a single impeller pattern, the total cost was 54 CNY. If the pattern is used for 50 castings, the cost per casting attributed to the pattern is only 1.08 CNY. In contrast, a wooden pattern costing 1,600 CNY would amortize to 32 CNY per casting over 50 uses. The 3D-printed pattern thus offers a 30-fold reduction in pattern cost per casting. Even if the pattern wears out after 10 cycles (PLA patterns can last for 20–50 cycles depending on handling), the cost per casting is still lower than that of a wooden pattern.
The success of this study demonstrates that FDM 3D printing is a viable alternative for producing patterns for sand casting of impellers and similar components. The combination of digital modeling and additive manufacturing enables rapid, cost-effective, and sustainable pattern production. This approach is particularly suited for small to medium batch sizes and complex geometries that are difficult to machine. Future work may involve optimizing the printing parameters for even better surface finish and dimensional accuracy, or exploring the use of composite filaments with higher strength and heat resistance to extend pattern life.
In summary, I have successfully replaced a traditional wooden impeller pattern with a PLA pattern manufactured by FDM 3D printing. The printed pattern was used in sand casting to produce aluminum impellers that met all quality requirements. The comparative analysis showed a dramatic improvement in material utilization (from 46% to 98%), a reduction in manufacturing cycle time (from 32 hours to 5 hours), and a cost reduction from 1,600 CNY to 54 CNY. The 3D-printed pattern also exhibited better durability and dimensional stability. This technology has great potential to transform the foundry industry by enabling faster and cheaper pattern production for sand casting.