In this research, I systematically investigated the process optimization of steel castings produced by 3D sand mold printing technology. The study combined response surface methodology (RSM) with numerical simulation using ProCAST to optimize both the sand mold printing parameters and the casting process design for a complex thin-walled stainless steel casing. The optimal printing parameters were determined as a layer thickness of 0.3 mm, furan resin content of 1.6 wt.% of sand, and curing agent content of 4‰ of sand, yielding a tensile strength of 1.055 MPa and gas evolution of 11.1 mL/g. Three different gating system designs were evaluated through filling and solidification simulation. The double-side injection method was selected as the best scheme, showing uniform filling and fewer defects. After adding risers and chills, the shrinkage porosity volume inside the casting was reduced from 0.391 cc to 0.0008 cc, demonstrating that the optimized casting process significantly improves the quality of steel castings. The study provides a comprehensive methodology for integrating 3D sand mold printing with casting simulation to produce high-quality steel castings with complex geometry.
1. Introduction
Additive manufacturing, commonly known as 3D printing, has revolutionized product development and manufacturing in various industries. Among the many branches of additive manufacturing, sand mold 3D printing is particularly promising for the foundry industry because it enables rapid production of sand molds and cores without the need for expensive tooling. Unlike traditional sand casting, where patterns and core boxes are required, sand mold 3D printing directly builds molds and cores layer by layer from digital CAD models. This approach offers high design freedom, short lead times, low cost for small batches, and the ability to produce complex geometries that are difficult or impossible to achieve with conventional methods.
For steel castings, the combination of 3D sand mold printing and advanced simulation tools has opened new possibilities for optimizing casting processes and improving product quality. Steel castings often have high melting points, poor fluidity, and a tendency to form shrinkage defects, making process design critical. Traditional trial-and-error methods are expensive and time-consuming. Numerical simulation software such as ProCAST can predict filling, solidification, and defect formation, allowing engineers to optimize gating systems, risers, chills, and process parameters before physical trials. When this is coupled with the geometric freedom of 3D sand printing, it becomes possible to design gating systems with shapes that minimize turbulence and heat loss, such as circular runners and tapered sprues, which are difficult to manufacture with conventional sand molding.
In this study, I aimed to develop a complete optimization framework for steel castings based on 3D sand mold printing. The work consists of two main parts: first, the optimization of the sand mold printing parameters using response surface methodology, and second, the design and simulation of the casting process for a natural gas meter housing made of 316 stainless steel. The goal was to identify the best combination of printing layer thickness, resin content, and curing agent content that maximizes tensile strength while minimizing gas evolution. Then, using the optimized sand mold parameters, I designed multiple gating system configurations and evaluated them through numerical simulation to select the one with the least defects. Finally, I further optimized the casting process by adding risers and chills to minimize shrinkage porosity. The results demonstrate that the integrated approach can significantly improve the quality of steel castings produced by sand mold 3D printing.
2. Materials and Methods
2.1 Raw Materials
The molding sand used in this study was a special silica sand designed for 3D printing applications. The properties are listed in Table 1. The sand has a uniform grain size, low acid demand, and low loss on ignition, making it suitable for resin bonding. The average particle size was about 100 mesh, and the shape was mostly spherical, which facilitates good packing and smooth layer spreading.
| Property | Value |
|---|---|
| Moisture content (%) | 0.08 |
| Clay content (%) | 0.12 |
| Acid demand value (%) | 3.1 |
| Loss on ignition (%) | 0.15 |
| Bulk density (g/cm³) | 1.47 |
The binder was a furan resin, whose properties are summarized in Table 2. Furan resin is widely used in sand mold 3D printing due to its excellent self-hardening characteristics and good collapsibility. The curing agent was a p-toluenesulfonic acid-based solution, with properties given in Table 3.
| Property | Value |
|---|---|
| Water content (%) | 3.47 |
| Nitrogen content (%) | 0.02 |
| Density (g/cm³) | 1.151 |
| Viscosity at 20°C (mPa·s) | 19 |
| Free formaldehyde (%) | 0.09 |
| Property | Value |
|---|---|
| Total acidity (%) | 0.08 |
| Density (g/cm³) | 0.12 |
| Free sulfuric acid (%) | 3.1 |
| Viscosity at 20°C (mPa·s) | 0.15 |
2.2 Sand Mold 3D Printing Equipment
The sand mold specimens and molds were produced using an industrial 3D sand printer with a build volume of 1000 mm × 800 mm × 600 mm. The printer uses a hopper system to spread sand mixed with curing agent, followed by jetting of furan resin through a printhead. The main controllable parameters were layer thickness (0.25–0.4 mm), resin content (0.5%–2% of sand mass), and curing agent content (2‰–4‰ of sand mass). The printing process includes mixing the sand and curing agent, spreading the mixed sand, jetting the resin, and then repeating the cycle until the build is completed. After printing, the molds were naturally cured for 24 hours at room temperature before testing.
2.3 Testing of Sand Specimens
To evaluate the mechanical and gas evolution properties of the printed sand specimens, standard “8-shaped” tensile specimens were printed according to GB/T 2684-2009. The tensile strength was measured using an XQY-II intelligent sand strength testing machine. The gas evolution was measured using a GET-III intelligent gas evolution tester. For gas testing, 1.00 g of powder was taken from the broken surface of tensile specimens and heated to 850°C in a quartz tube, and the gas volume was recorded. Each test was repeated at least six times to obtain an average value.
3. Response Surface Optimization of Sand Mold 3D Printing Parameters
3.1 Single-Factor Experiments
Before applying response surface methodology, single-factor experiments were carried out to determine the central values and the influence ranges of the key process parameters. The three parameters investigated were printing layer thickness, furan resin content, and curing agent content. The response variables were tensile strength and gas evolution of the sand mold.
3.1.1 Effect of Layer Thickness
In this set of experiments, the resin content was fixed at 2% of sand mass, the curing agent content at 3.5‰ of sand mass, and the layer thickness was varied from 0.30 mm to 0.40 mm in steps of 0.05 mm. The results are shown in Table 4. As the layer thickness increased, the tensile strength decreased from 0.993 MPa to 0.626 MPa, while the gas evolution decreased from 12.9 mL/g to 10.8 mL/g. This occurs because a thicker layer reduces the overall resin content in the mold (since fewer layers are printed) and also reduces the packing density, weakening the bond bridges between sand grains. The lower gas evolution is attributed to the decreased amount of organic resin in the sand.
| Layer Thickness (mm) | Tensile Strength (MPa) | Gas Evolution (mL/g) |
|---|---|---|
| 0.30 | 0.993 | 12.9 |
| 0.35 | 0.832 | 11.8 |
| 0.40 | 0.626 | 10.8 |
3.1.2 Effect of Resin Content
With the layer thickness fixed at 0.35 mm and curing agent at 3.5‰, the resin content was varied from 1.0% to 2.0% of sand mass. The results are presented in Table 5. The tensile strength increased from 0.714 MPa to 1.011 MPa as the resin content increased, while the gas evolution also increased from 10.1 mL/g to 11.9 mL/g. Higher resin content creates more bonding bridges and improves the tensile strength, but it also introduces more organic material that decomposes to gas at high temperatures.
| Resin Content (%) | Tensile Strength (MPa) | Gas Evolution (mL/g) |
|---|---|---|
| 1.0 | 0.714 | 10.1 |
| 1.5 | 0.871 | 11.0 |
| 2.0 | 1.011 | 11.9 |
3.1.3 Effect of Curing Agent Content
Fixing the layer thickness at 0.35 mm and resin content at 1.5%, the curing agent content was changed from 3.0‰ to 4.0‰. Table 6 shows that the tensile strength increased from 0.715 MPa to 0.731 MPa when the curing agent increased from 3.0‰ to 4.0‰. The gas evolution increased from 8.3 mL/g to 12.2 mL/g over the same range. Although more curing agent promotes a more complete polymerization reaction, excessive curing agent may also introduce sulfur-containing gases, increasing gas evolution. Therefore, a moderate curing agent content around 3.5‰ was chosen as the central value for response surface experiments.
| Curing Agent Content (‰) | Tensile Strength (MPa) | Gas Evolution (mL/g) |
|---|---|---|
| 3.0 | 0.715 | 8.3 |
| 3.5 | 0.728 | 11.2 |
| 4.0 | 0.731 | 12.2 |
3.2 Response Surface Experiment Design
Based on the single-factor results, the central values were chosen as: layer thickness = 0.35 mm, resin content = 1.5%, and curing agent content = 3.5‰. A three-factor, three-level Box-Behnken design was adopted. The three independent variables were: A: layer thickness (0.3, 0.35, 0.4 mm); B: resin content (1.0%, 1.5%, 2.0%); C: curing agent content (3.0‰, 3.5‰, 4.0‰). The responses were tensile strength (Y₁) and gas evolution (Y₂). Table 7 lists the experimental runs and measured responses.
| Run | A (mm) | B (%) | C (‰) | Tensile Strength (MPa) | Gas Evolution (mL/g) |
|---|---|---|---|---|---|
| 1 | 0.3 | 1.0 | 3.5 | 0.682 | 10.8 |
| 2 | 0.4 | 1.0 | 3.5 | 0.535 | 9.4 |
| 3 | 0.3 | 2.0 | 3.5 | 1.071 | 13.6 |
| 4 | 0.4 | 2.0 | 3.5 | 0.518 | 12.3 |
| 5 | 0.3 | 1.5 | 3.0 | 0.712 | 8.8 |
| 6 | 0.4 | 1.5 | 3.0 | 0.533 | 7.8 |
| 7 | 0.3 | 1.5 | 4.0 | 1.011 | 14.2 |
| 8 | 0.4 | 1.5 | 4.0 | 0.468 | 11.9 |
| 9 | 0.35 | 1.0 | 3.0 | 0.426 | 8.6 |
| 10 | 0.35 | 2.0 | 3.0 | 0.586 | 8.1 |
| 11 | 0.35 | 1.0 | 4.0 | 0.679 | 11.8 |
| 12 | 0.35 | 1.5 | 3.5 | 0.754 | 9.9 |
| 13 | 0.35 | 1.5 | 3.5 | 0.714 | 11.9 |
| 14 | 0.35 | 1.5 | 3.5 | 0.634 | 12.7 |
| 15 | 0.35 | 1.5 | 3.5 | 0.628 | 11.9 |
| 16 | 0.35 | 1.5 | 3.5 | 0.681 | 12.4 |
| 17 | 0.35 | 1.5 | 3.5 | 0.637 | 11.5 |
3.3 Regression Models and ANOVA
Using Design-Expert software, second-order polynomial models were fitted to the experimental data. The regression equation for tensile strength (Y₁) is:
$$Y_1 = -0.66 – 0.18A + 0.076B + 0.082C – 0.10AB – 0.091AC – 0.021BC + 0.056A^2 – 0.014B^2 – 0.034C^2 \tag{1}$$
The regression equation for gas evolution (Y₂) is:
$$Y_2 = 10.72 – 1.08A + 1.21B + 0.36C + 0.075AB + 0.025AC – 1.05BC – 0.54A^2 + 0.44B^2 – 0.61C^2 \tag{2}$$
Tables 8 and 9 present the ANOVA results for the two models. For tensile strength, the model F-value is 28.01 (p = 0.0001), indicating high significance. The lack-of-fit is not significant (p = 0.3212), meaning the model adequately fits the data. The determination coefficient R² = 0.9730 and adjusted R² = 0.9382 are both greater than 0.9, confirming excellent predictive ability. Among the individual factors, A (layer thickness) has the largest influence (F = 142.67), followed by C (curing agent) and B (resin content). The interactions AB and AC are significant, while BC is not.
| Source | Sum of Squares | df | Mean Square | F Value | p-value | Significance |
|---|---|---|---|---|---|---|
| Model | 0.45 | 9 | 0.050 | 28.01 | 0.0001 | ** |
| A | 0.25 | 1 | 0.25 | 142.67 | <0.0001 | ** |
| B | 0.046 | 1 | 0.046 | 26.00 | 0.0014 | ** |
| C | 0.054 | 1 | 0.054 | 30.27 | 0.0009 | ** |
| AB | 0.041 | 1 | 0.041 | 23.26 | 0.0019 | ** |
| AC | 0.033 | 1 | 0.033 | 18.70 | 0.0035 | ** |
| BC | 1.81e-4 | 1 | 1.81e-4 | 1.02 | 0.3463 | – |
| A² | 0.016 | 1 | 0.013 | 7.49 | 0.0291 | * |
| B² | 7.82e-4 | 1 | 1.81e-4 | 1.12 | 0.5278 | – |
| C² | 4.90e-4 | 1 | 4.90e-3 | 2.77 | 0.1401 | – |
| Residual | 0.012 | 7 | 1.77e-3 | |||
| Lack of Fit | 6.77e-3 | 3 | 2.26e-3 | 1.61 | 0.3212 | not significant |
| Pure Error | 5.62e-3 | 4 | 1.41e-4 | |||
| Cor Total | 0.46 | 16 |
| Source | Sum of Squares | df | Mean Square | F Value | p-value | Significance |
|---|---|---|---|---|---|---|
| Model | 30.02 | 9 | 3.34 | 68.57 | <0.0001 | ** |
| A | 9.24 | 1 | 9.24 | 190.06 | <0.0001 | ** |
| B | 11.76 | 1 | 11.76 | 241.79 | <0.0001 | ** |
| C | 1.05 | 1 | 1.05 | 21.61 | 0.0023 | ** |
| AB | 0.023 | 1 | 0.023 | 0.46 | 0.5183 | – |
| AC | 2.50e-3 | 1 | 2.50e-3 | 0.051 | 0.8271 | – |
| BC | 4.41 | 1 | 4.41 | 90.66 | <0.0001 | ** |
| A² | 1.21 | 1 | 1.21 | 24.78 | 0.0016 | * |
| B² | 0.82 | 1 | 0.82 | 16.76 | 0.0046 | * |
| C² | 1.57 | 1 | 1.57 | 32.21 | 0.0008 | ** |
| Residual | 0.34 | 7 | 0.049 | |||
| Lack of Fit | 0.23 | 3 | 0.077 | 2.87 | 0.1673 | not significant |
| Pure Error | 0.11 | 4 | 0.027 | |||
| Cor Total | 30.86 | 16 |
For gas evolution, the model is also highly significant with R² = 0.9888 and adjusted R² = 0.9744. The lack-of-fit is not significant (p = 0.1673). The order of influence on gas evolution is B > A > C, meaning resin content has the largest effect, followed by layer thickness and curing agent. The interaction BC is significant, meaning resin and curing agent jointly affect gas evolution, while AB and AC are not.
3.4 Interaction Effects and Optimization
Response surface plots were used to visualize the interaction effects. For tensile strength, the interactions of layer thickness with resin content and curing agent content showed significant curvature, confirming their significance. For gas evolution, the interaction between resin and curing agent exhibited a pronounced curvature, validating the ANOVA result. Based on the desirability function, the optimum parameters for maximizing tensile strength and minimizing gas evolution were found to be: layer thickness = 0.3 mm, resin content = 1.629%, and curing agent content = 4.0‰. Considering practical limitations, the parameters were rounded to 0.3 mm, 1.6%, and 4.0‰, respectively. Verification experiments were carried out with these parameters, and the results are shown in Table 10. The average tensile strength was 1.055 MPa, and the average gas evolution was 11.1 mL/g, which closely matches the predicted values, confirming the reliability of the model.
| No. | Layer Thickness (mm) | Resin (%) | Curing Agent (‰) | Tensile Strength (MPa) | Gas Evolution (mL/g) |
|---|---|---|---|---|---|
| 1 | 0.3 | 1.6 | 4 | 1.063 | 11.0 |
| 2 | 0.3 | 1.6 | 4 | 1.049 | 11.3 |
| 3 | 0.3 | 1.6 | 4 | 1.050 | 11.2 |
| 4 | 0.3 | 1.6 | 4 | 1.058 | 10.9 |
| Average | 1.055 | 11.1 |
4. Casting Process Design for Steel Castings Using 3D Printed Sand Mold
4.1 Component Description and Casting Characteristics
The object of this study is a natural gas meter housing made of 316 stainless steel (equivalent to 0Cr17Ni12Mo2). This component has a complex thin-walled structure with a central partition plate containing four holes, three bosses on the side, and a flange section. The wall thickness varies from 4 mm to 15 mm. The casting requires high strength, pressure tightness, and corrosion resistance. Because of the thin walls and uneven thickness distribution, conventional sand casting would require multiple cores and careful gating design. The use of 3D sand printing allows the mold to be manufactured as a single piece or with fewer cores, reducing assembly errors and enabling more complex runner geometries.
4.2 Gating System Design
Considering the poor fluidity and high melting point of 316 stainless steel, a closed-open gating system was selected. Circular cross-sections were used for the runners because they provide lower heat loss and better flow behavior than rectangular sections. The gating system was designed using the section-ratio method. For this thin-walled steel casting, the cross-sectional area ratio of the sprue (A直), runner (A横), and ingate (A内) was chosen as:
$$A_{\text{直}} : A_{\text{横}} : A_{\text{内}} = 11 : 9 : 10 \tag{3}$$
The total cross-sectional area of the ingates was calculated using the following formula:
$$A_{\text{内}} = \frac{G}{\mu \rho \tau \sqrt{2g H_p}} \tag{4}$$
where G is the mass of metal flowing through the gating system, ρ is the density of the molten metal, τ is the total pouring time, μ is the flow coefficient, and Hp is the average pressure head. For a middle-injection system, Hp is given by:
$$H_p = H_0 – 0.5p \tag{5}$$
where H0 is the total height of the sprue and p is the height of the casting above the ingate. Calculations yielded the final dimensions: sprue radius = 15 mm, runner radius = 12 mm, ingate radius = 13 mm. These dimensions ensure a reasonable filling rate and reduce turbulence.
4.3 Three Proposed Gating Designs
To find the best gating configuration, I designed three different schemes as shown in Table 11. Scheme A used a single-side injection with only one ingate. Scheme B used double-side injection with two ingates placed symmetrically. Scheme C used reverse double-side injection, where the ingates were placed so that the metal enters in opposite directions, intended to promote earlier solidification of the flange area.
| Scheme | Description | Advantages | Potential Issues |
|---|---|---|---|
| A | Single-side injection | Simplest design, fast filling | Non-uniform filling, risk of cold shuts, more defects |
| B | Double-side injection | Uniform filling, synchronized solidification | More complex runner |
| C | Reverse double-side injection | Can promote early flange solidification | Turbulence at junction, slower filling, isolated liquid regions |
4.4 Numerical Simulation Setup
The three gating designs were modeled in NX CAD software and exported as Parasolid files for ProCAST simulation. The mesh was generated with triangular surface elements. A global mesh size of 3 mm was used for the casting, and 4 mm for the sand mold. Local refinement was applied at thin-wall regions and complex features to ensure accurate representation. The material properties of 316 stainless steel were defined in ProCAST based on its chemical composition and thermodynamic calculation. Table 12 lists the chemical composition and mechanical properties of the selected stainless steel.
| Element/Property | Value |
|---|---|
| C (%) | ≤0.08 |
| Si (%) | ≤1.00 |
| Mn (%) | ≤2.00 |
| P (%) | ≤0.035 |
| S (%) | ≤0.03 |
| Ni (%) | 10–14 |
| Cr (%) | 16.0–18.5 |
| Mo (%) | 2.0–3.0 |
| Tensile strength (MPa) | ≥520 |
| Yield strength (MPa) | ≥205 |
| Elongation (%) | ≥40 |
| Reduction of area (%) | ≥60 |
| Density (g/cm³) | 7.98 |
| Specific heat at 20°C (J/g·K) | 0.502 |
Figure 5 shows the fraction solid, enthalpy, and density of the steel as functions of temperature. The liquidus temperature is approximately 1505°C. The pouring temperature was set to 1560°C to account for heat losses during filling. The sand mold was defined as furan resin sand with the optimized printing parameters from the previous section. The initial mold temperature was 25°C. Heat transfer coefficients were assigned as: casting–mold: 1000 W/(m²·K), casting–chill: 2000 W/(m²·K), mold–air: convection. The filling time was calculated to be 4 seconds, and the total mass of metal poured was 4.237 kg.
4.5 Filling Simulation Results
The filling process for the three schemes was simulated. Scheme A exhibited a fast but non-uniform filling. At t = 1 s, the metal was concentrated on the right side near the ingate, while the left side remained empty. This uneven front can lead to cold shuts and entrapped air. At t = 3 s, a height difference existed between the two sides, increasing the risk of oxide inclusion. Scheme B produced a smooth and symmetric filling. The metal entered from both sides simultaneously, and the melt fronts rose at almost the same rate. There was minor turbulence at the junction where the two streams met, but no severe entrapment. Scheme C started by filling the flange area first, causing some turbulent flow in the bosses. The filling was slower, and at the end of 4 s there was still an unfilled region in the upper part of the casting, indicating a risk of misrun.
4.6 Solidification and Temperature Field Analysis
The solidification time and temperature distribution for the three schemes are presented in Table 13. Scheme A displayed a large variation in solidification time, with a hot spot in the lower thick section that was isolated by already solidified thin walls. This could lead to shrinkage cavities and porosity. Scheme B had a more uniform solidification pattern, following a directional solidification from the bottom to the top. The thick flange areas took longer to solidify (about 71 s), while the thin central web solidified in about 9 s. Scheme C exhibited overall slower solidification, and the region near the runner remained hot even after 140 s, indicating poor feeding and a high chance of shrinkage porosity.
| Scheme | Solidification Pattern | Hot Spot Location | Defect Risk |
|---|---|---|---|
| A | Non-uniform, isolated liquid pool | Lower thick plate and flange | High |
| B | Quasi-directional, uniform | Flange bosses | Moderate |
| C | Slow, irregular | Runner side and lower thin plate | High |
4.7 Defect Prediction
The shrinkage porosity was predicted using the Niyama criterion. The Niyama parameter is defined as:
$$M = \frac{G}{\sqrt{R}} \tag{6}$$
where G is the temperature gradient and R is the cooling rate. When M is less than or equal to 1.0, the local region is highly prone to shrinkage porosity. Table 14 summarizes the predicted defect volume for each scheme. Scheme A had a total defect volume of 0.559 cc, located mainly in the lower separator and the right side of the flange. Scheme B had 0.391 cc of defects, concentrated in the flange boss areas, which could be reduced by risers. Scheme C had 0.499 cc of dispersed defects, including some on critical surfaces. Based on the simulation, Scheme B was selected for further optimization because it offered the best combination of filling behavior, solidification pattern, and defect distribution.
| Scheme | Total Defect Volume (cc) | Shrinkage Porosity Volume (cc) | Density (g/cc) | Weight (mg) |
|---|---|---|---|---|
| A | 3.533 | 0.559 | 1.20 | 0.671 |
| B | 3.445 | 0.391 | 1.20 | 0.469 |
| C | 2.476 | 0.499 | 1.20 | 0.599 |
5. Process Optimization of the Double-Side Injection Scheme
5.1 Riser Design
To eliminate the shrinkage defects in the flange boss areas, two open top risers were designed on the top of the flange. The riser dimensions were calculated using the modulus method. The modulus of the riser is defined as:
$$M_r = \frac{V_r}{A_r} \tag{7}$$
where Vr is the riser volume and Ar is the cooling surface area. The calculated riser modulus was 3.8 mm. Using the hot spot circle method, the riser diameter was determined to be 29 mm, and the height was calculated as H = 1.15–1.8D, giving H = 35 mm. A neck was formed at the bottom of the riser, and both upper and lower edges were rounded with a radius of 4 mm. The riser geometry is shown in Table 15.
| Parameter | Value (mm) |
|---|---|
| Diameter | 29 |
| Height | 35 |
| Top fillet radius | 4 |
| Bottom fillet radius | 4 |
| Neck diameter | 16 (approximately) |
5.2 Chill Design
The lower thick section of the casting remained a potential hot spot. To promote directional solidification, an external steel chill was placed under this thick region. The chill dimensions were chosen to increase the cooling rate locally without causing excessive thermal stress. The chill is approximately 25 mm × 25 mm × 60 mm, matching the area of the thick section. The chill material was set as iron, and the interfacial heat transfer coefficient between the chill and the casting was set to 2000 W/(m²·K).
5.3 Casting Process Parameters
Based on the casting manual and the dimensions of the component, the casting tolerance grade was selected as CT13 with a tolerance size of 10 mm. The weight tolerance grade was M13, corresponding to a weight tolerance value of 24%. The machining allowance was determined to be 9 mm for the machining grade 13/J. For the flange area, a process correction allowance of 2 mm was added to compensate for possible distortion during solidification.
5.4 Optimized Simulation Results
The optimized casting process (Scheme B plus risers and chill) was simulated again. The filling process became even more uniform than the original Scheme B. The melt entered the cavity through both ingates at a velocity of about 0.7 m/s, then slowed to 0.5 m/s as it filled the lower thick region. No height difference appeared between the two sides, and all parts were filled completely within 4 seconds.
The solidification analysis showed that the chill successfully accelerated the cooling of the lower thick section, creating a favorable bottom-to-top directional solidification. The risers effectively fed the flange bosses, and no isolated liquid pools were observed. The overall solidification time was slightly reduced compared with the unoptimized scheme.
The final defect distribution after optimization is given in Table 16 and visually represented in Figure 21 in the original thesis. The shrinkage porosity inside the casting was only 0.0008 cc, located in a non-critical area, while the riser contained 0.073 cc of defects. This represents a dramatic improvement compared with the original 0.391 cc of defects. The results confirm that the optimized process design is effective for producing sound steel castings.
| Condition | Shrinkage Porosity Volume (cc) |
|---|---|
| Original Scheme B (without risers/chills) | 0.391 |
| After adding risers and chill | 0.0008 (in casting) |
| Defects in risers | 0.073 |
6. Conclusions
In this work, I conducted a comprehensive study on the process optimization of steel castings based on 3D sand mold printing. The main findings and conclusions are as follows:
- The response surface methodology was successfully applied to optimize the 3D sand printing parameters. The optimal combination was a layer thickness of 0.3 mm, a furan resin content of 1.6% by mass of sand, and a curing agent content of 4‰ by mass of sand. The verification experiments produced a tensile strength of 1.055 MPa and a gas evolution of 11.1 mL/g, which agreed well with the model predictions.
- For tensile strength, the order of influencing factors was layer thickness > curing agent content > resin content. For gas evolution, the order was resin content > layer thickness > curing agent content. The interaction between resin content and curing agent content had a significant effect on gas evolution.
- A closed-open gating system with circular cross-section runners was designed for a natural gas meter housing of 316 stainless steel. The section ratio of sprue: runner: ingate was 11:9:10, and three gating schemes were compared numerically.
- The double-side injection scheme was selected as the best among the three because it provided the most uniform filling and solidification and the lowest defect volume (0.391 cc). The single-side injection produced non-uniform filling, while the reverse double-side injection caused turbulence and slower solidification.
- With the addition of two open risers and an external chill, the shrinkage porosity inside the cast part was reduced from 0.391 cc to 0.0008 cc, with most remaining defects located in the risers (0.073 cc). The optimized process eliminated the hot spot in the lower thick section and established a directional solidification pattern.
- The integration of 3D sand mold printing with numerical simulation provides a powerful approach for producing high-quality steel castings with complex thin-walled geometries, reducing development time and cost.
This study demonstrates that sand mold 3D printing combined with response surface optimization and ProCAST simulation can be used to systematically design and optimize casting processes for steel castings. The methodology can be extended to other materials and components, offering a modern and sustainable route for foundry production.