In the manufacturing of heavy-duty components for sectors such as mining, transportation, and heavy machinery, the production of high-integrity steel castings remains a critical challenge. Among various methods, sand casting is extensively utilized for fabricating complex sand casting parts due to its flexibility, cost-effectiveness for low to medium volumes, and ability to produce large components. However, the process is inherently prone to defects like gas porosity, inclusions, mistuns, shrinkage cavities, and macro-porosity, which significantly compromise the quality and yield of the final product. The goal of my research is to systematically address these issues by leveraging numerical simulation to design, analyze, and optimize the sand casting process for a specific steel shell component. By virtually testing different gating systems and process parameters, my study aims to minimize defect formation, enhance the qualification rate, and establish a reliable methodological framework applicable to similar sand casting parts.
The subject of this investigation is a medium-carbon cast steel (ZG270-500) shell, a representative component with demanding quality requirements. The chemical composition of the alloy used is detailed in Table 1. This composition provides a good balance of strength, hardness, and a measure of toughness, making it suitable for components bearing significant loads. The microstructure typically consists of ferrite and pearlite, offering favorable machinability. The geometry of the shell presents typical challenges encountered in sand casting parts: it is hollow, has a varying wall thickness with some bulky sections, and features internal surfaces requiring high quality. The casting has a mass of approximately 392.93 kg and overall dimensions of 812 mm × 525 mm × 356 mm, with an average wall thickness of 8 mm.
| Element | C | Mn | Si | P (max) | S (max) | Fe |
|---|---|---|---|---|---|---|
| Content | 0.4 – 0.5 | 0.7 – 0.8 | 0.2 – 0.45 | 0.04 | 0.05 | Bal. |
Based on the geometry and quality requirements, three distinct sand casting process schemes were conceived. The fundamental design principles focused on selecting an appropriate parting plane and designing a gating system that promotes smooth filling and effective feeding. For all schemes, an open gating system was chosen to ensure non-turbulent, steady filling, thereby reducing the risk of oxide inclusion formation—a common issue in steel sand casting parts. The pouring time and flow rates were calculated based on standard foundry handbook formulas to establish a baseline for simulation.
The pouring time $t$ (s) is calculated from the total metal mass $G_L$ (kg), the number of ladles $N$, the number of pouring holes per ladle $n$, and the average pouring rate $q$ (kg/s):
$$ t = \frac{G_L}{N \cdot n \cdot q} $$
For this casting, with $G_L = 450$ kg (including gating system), $N=1$, $n=1$, and $q = 27$ kg/s, the calculated pouring time was $t \approx 16.7$ s.
This time was verified by checking the upward velocity $v$ (mm/s) of the metal in the mold cavity, which must be sufficient to prevent mistuns. The velocity is given by the height of the casting $C$ (mm) divided by the pouring time:
$$ v = \frac{C}{t} $$
With $C = 356$ mm, the resulting upward velocity was $v \approx 21.3$ mm/s, which is generally acceptable for steel castings of this size.
The initial gating system cross-sectional areas were determined using standard ratios for steel castings. With a bottom-pour ladle nozzle diameter of 40 mm (area $A_{nozzle} \approx 1256 \text{ mm}^2$), the choke area $\sum A_{choke}$ was set equal to the nozzle area. The subsequent areas for the sprue ($\sum A_{sprue}$), runners ($\sum A_{runner}$), and ingates ($\sum A_{ingate}$) were scaled accordingly using the ratio:
$$ \sum A_{choke} : \sum A_{sprue} : \sum A_{runner} : \sum A_{ingate} = 1 : (1.8-2.0) : (1.8-2.0) : (2.0-2.5) $$
The core characteristics of the three designed schemes are summarized below:
| Scheme | Ingate Location | Risers | Primary Design Rationale |
|---|---|---|---|
| Scheme 1 | At the base flange of the shell | None initially, later 2 side risers added | Simple molding, bottom filling to reduce turbulence. |
| Scheme 2 | On the cylindrical body of the shell | None initially, later 2 side risers added | Direct feeding into the main body section. |
| Scheme 3 (Optimized) | Modified location with multiple ingates | 4 risers (2 cylindrical, 2 stepped) | Comprehensive feeding for all hot spots and directional solidification. |
Numerical simulation was performed using ProCAST software, a powerful tool for modeling filling, solidification, and defect formation in casting processes. A critical pre-processing step was the generation of a high-quality mesh. For the shell casting and its gating system, a mesh size of 30 mm was selected, resulting in 12,722 surface elements and 45,906 volume elements. The material for the casting was defined as Medium-Carbon AISI 1040 (equivalent to ZG270-500), and the mold and cores were defined as silica sand bonded with resin. Key boundary conditions included a mold initial temperature of 25°C, interfacial heat transfer coefficients of 1000 W/(m²·K), and gravitational acceleration.
The initial simulations for Scheme 1 and Scheme 2, with a pouring temperature of 1560°C and a pouring velocity of 1.6 m/s, revealed a critical flaw: both schemes resulted in a mistun (incomplete filling) at the top of the cylindrical section of the shell. This immediately disqualified the basic versions of these schemes. To address this, two side risers were added to each scheme in the under-filled region. Subsequent simulation of the solidification and shrinkage porosity for these modified schemes, however, showed unacceptable levels of internal defects. Scheme 1 exhibited a total shrinkage porosity volume of 28.23 cm³, while Scheme 2 was worse at 50.58 cm³. The defects were concentrated in the thicker sections and junctions, indicating inadequate feeding. These results underscored that simply adding risers to an ill-conceived gating layout was insufficient for producing sound sand casting parts.
The analysis of the first two schemes informed the design of a comprehensively optimized Scheme 3. The key modifications were: 1) Re-routing the ingate system to ensure balanced filling and to position the metal entry points more strategically relative to the bulky sections. 2) Incorporating a robust risering system consisting of four open risers. Two were cylindrical (Ø110 mm x 350 mm high), and two were strategically placed stepped risers (lower section Ø200 mm x 32.5 mm, upper section Ø135 mm x 330 mm) to effectively feed the major hot spots identified in previous simulations.
The filling sequence for Scheme 3, as shown by simulation, was significantly improved. The metal entered the mold cavity smoothly and filled it progressively from the bottom upwards, with no isolated liquid pockets or premature freezing. The temperature distribution during filling was favorable, with the thinner base sections cooling faster and the thicker cylindrical sections retaining heat longer—a desirable pattern for directional solidification. The solidification simulation confirmed this pattern. The thinner sections and areas near the ingates solidified first, creating a thermal gradient that pushed the final liquid metal towards the risers. The solidification fronts moved sequentially from the casting extremities towards the risers, which remained liquid longest. This is the classic condition for achieving sound, dense sand casting parts, as the risers can effectively compensate for the volumetric shrinkage of solidifying steel.
To quantitatively determine the optimal operating parameters within the optimized Scheme 3 geometry, a virtual Design of Experiments (DoE) was conducted. The two most influential process parameters for defect formation in sand casting—pouring temperature and pouring velocity—were selected as factors. A two-factor, three-level orthogonal array (L9) was designed, as shown in Table 3. The response variable was the total predicted volume of shrinkage porosity and cavities from the simulation.
| Level | Factor A: Pouring Temperature (°C) | Factor B: Pouring Velocity (m/s) |
|---|---|---|
| 1 | 1530 | 1.3 |
| 2 | 1560 | 1.6 |
| 3 | 1590 | 1.9 |
Nine separate simulations were run according to the orthogonal array. The results for the shrinkage porosity volume are listed in Table 4. Range analysis (R) was then performed on this data. For each factor, the average response (K) for each level was calculated. The range R for a factor is the difference between the maximum and minimum of these level averages. A larger R value indicates a greater influence of that factor on the response.
| Run No. | A: Temp. (°C) | B: Velocity (m/s) | Porosity Volume (cm³) |
|---|---|---|---|
| 1 | 1530 | 1.3 | 2.368 |
| 2 | 1530 | 1.6 | 2.201 |
| 3 | 1530 | 1.9 | 2.503 |
| 4 | 1560 | 1.3 | 1.553 |
| 5 | 1560 | 1.6 | 1.416 |
| 6 | 1560 | 1.9 | 1.818 |
| 7 | 1590 | 1.3 | 1.984 |
| 8 | 1590 | 1.6 | 2.066 |
| 9 | 1590 | 1.9 | 2.206 |
Level Averages (K):
For Temperature (A): $K_{A1} = (2.368+2.201+2.503)/3 = 2.357$; $K_{A2} = (1.553+1.416+1.818)/3 = 1.596$; $K_{A3} = (1.984+2.066+2.206)/3 = 2.085$.
For Velocity (B): $K_{B1} = (2.368+1.553+1.984)/3 = 1.968$; $K_{B2} = (2.201+1.416+2.066)/3 = 1.894$; $K_{B3} = (2.503+1.818+2.206)/3 = 2.176$.
Range (R) Calculation:
$R_A = max(2.357, 1.596, 2.085) – min(2.357, 1.596, 2.085) = 2.357 – 1.596 = 0.761$ (in the context of the level averages).
$R_B = max(1.968, 1.894, 2.176) – min(1.968, 1.894, 2.176) = 2.176 – 1.894 = 0.282$.
A more illustrative comparison for the main effect can be seen from the totals (sum) for each level across the three runs, as often used in orthogonal analysis: $\sum K_{A1} = 7.072$, $\sum K_{A2} = 4.787$, $\sum K_{A3} = 6.256$. The range based on these sums is $R_A = 7.072 – 4.787 = 2.285$. Similarly for B: $\sum K_{B1} = 5.905$, $\sum K_{B2} = 5.683$, $\sum K_{B3} = 6.527$, $R_B = 6.527 – 5.683 = 0.844$. This clearly shows $R_A > R_B$.
The analysis leads to several key conclusions. First, pouring temperature (Factor A) has a more significant influence on shrinkage porosity volume than pouring velocity (Factor B), as evidenced by its larger range value. Second, the optimal level for each factor is the one yielding the smallest level average (or sum) for porosity volume. For temperature, Level 2 (1560°C) is optimal ($K_{A2}=1.596$). For velocity, Level 2 (1.6 m/s) is optimal ($K_{B2}=1.894$). Therefore, the optimal parameter combination is A2B2: a pouring temperature of 1560°C and a pouring velocity of 1.6 m/s. Simulation under these conditions within the optimized Scheme 3 geometry confirmed a minimal shrinkage porosity volume of only 1.416 cm³, which is a drastic reduction compared to the initial schemes.
The final optimized sand casting process—Scheme 3 with parameters of 1560°C and 1.6 m/s—was validated through practical production. The casting yield (or pour weight ratio) for this optimized process was calculated to be approximately 66%. Most importantly, the qualification rate of the produced shell castings increased dramatically from an initial baseline of around 81% to 96%. The internal soundness and mechanical properties of the castings met all specified requirements, demonstrating the effectiveness of the simulation-driven optimization.
In summary, my research demonstrates a systematic approach to optimizing the sand casting process for complex steel components. The key findings are: 1) Initial schemes based on conventional layout failed due to mistuns and significant shrinkage, highlighting the limitations of trial-and-error methods for complex sand casting parts. 2) A holistic redesign focusing on controlled filling and directional solidification via strategic gating and risering (Scheme 3) was essential. 3) Numerical simulation enabled precise visualization of filling patterns and solidification sequences, guiding these design choices effectively. 4) Orthogonal experimentation via simulation identified the optimal combination of pouring temperature (1560°C) and velocity (1.6 m/s), minimizing shrinkage defects. 5) The validated process achieved a high qualification rate (96%), proving the methodology’s success. This workflow—integrating geometric design, multi-scenario simulation, and statistical parameter optimization—provides a powerful, cost-effective framework for developing robust processes for a wide range of demanding sand casting parts, reducing lead times, minimizing scrap, and improving economic outcomes in foundry operations.