In modern foundry practices, the integration of computer-aided engineering (CAE) with experimental techniques has revolutionized process optimization, particularly in lost foam casting, also known as expendable pattern casting (EPC). This method, which utilizes foam patterns that vaporize during metal pouring, offers significant cost advantages for small-batch production by eliminating the need for expensive molds. However, challenges such as gas generation and residue accumulation necessitate precise control over thermal parameters to ensure casting quality. In this study, we combine CAE simulations with temperature measurement tests to optimize thermophysical properties, enhance simulation accuracy, and explore the relationship between cooling rates and microstructural evolution in thick-section ductile iron castings. Our approach focuses on improving the reliability of numerical predictions for defects like shrinkage porosity and providing insights into pearlite content control, all within the context of lost foam casting processes.
The foundation of our research lies in the critical role of temperature field accuracy in CAE simulations. Parameters such as thermal impedance, specific heat capacity, and thermal conductivity significantly influence the computed results, including cooling rates and defect predictions. By employing inverse calculation capabilities in ADSTEFAN software, we refine these parameters based on experimental temperature curves obtained at the casting-mold interface. This optimization not only aligns simulation outcomes with real-world conditions but also establishes a framework for advancing EPC methodologies. Furthermore, we utilize numerical analysis to derive functional relationships between temperature profiles and cooling rates, enabling predictive control over microstructural features like pearlite formation. Throughout this work, we emphasize the practical applicability of our findings to industrial settings, underscoring the versatility of lost foam casting for complex geometries.
Experimental Setup and Temperature Measurement Methodology
Our investigation centers on a automotive stamping die casting made of GGG70L ductile iron, with external dimensions of 1500 mm × 2400 mm × 300 mm and a weight of 2700 kg. This thick-section casting exemplifies the typical challenges in lost foam casting, where uneven cooling can lead to defects. The pouring temperature was maintained at 1415 ± 2 °C, and we used self-curing furan resin sand for the EPC process. Cooling was monitored over 50 hours to capture the entire solidification profile. To address the harsh conditions during pouring—such as high temperatures and chemical reactions—we selected K-type thermocouples without ceramic protection tubes for enhanced sensitivity. These thermocouples were positioned 2–5 mm from the foam pattern surface within the sand mold, avoiding direct contact with the casting to prevent damage while ensuring accurate reflection of interfacial temperatures.
We conducted two primary experiments to validate our approach. In the first experiment, thermocouples were placed at strategic points around the casting, as illustrated in the setup diagram. The initial thermophysical parameters for GGG70L and furan resin sand, along with thermal resistance values, are summarized in Tables 1 and 2. These parameters served as the baseline for CAE simulations using ADSTEFAN. The second experiment introduced chill elements (made of HT300) to examine their impact on temperature distribution, with four thermocouples deployed across two castings in the same flask. This design allowed us to assess the consistency of optimized parameters under varied conditions, reinforcing the relevance of our findings to real-world lost foam casting applications.
| Parameter | GGG70L | Furan Resin Sand |
|---|---|---|
| Liquidus Temperature (°C) | 1150 | – |
| Solidus Temperature (°C) | 1145 | – |
| Latent Heat of Crystallization (J/g) | 50 | – |
| Thermal Conductivity (W/cm·°C) | 0.05 | 0.0025 |
| Specific Heat Capacity (J/g·°C) | 0.17 | 0.25 |
| Density (g/cm³) | 7.2 | 1.6 |
| Interface | Thermal Resistance (cm²·°C·s/J) |
|---|---|
| Chill-Sand | 5 |
| Chill-Casting | 30 |
| Casting-Sand | 50 |
The temperature data collected from these experiments revealed distinct curves characterized by rapid initial heating followed by gradual cooling. Comparative analysis with ADSTEFAN simulations showed significant discrepancies, with errors exceeding 200 °C in some regions. This highlighted the need for parameter optimization to improve simulation fidelity. Using the inverse calculation module, we iteratively adjusted the thermophysical properties and thermal resistances until the standard deviation between experimental and simulated curves dropped from 271 to 30. The optimized parameters, detailed in Tables 3 and 4, formed the basis for subsequent analyses and validations.
| Parameter | GGG70L | Furan Resin Sand |
|---|---|---|
| Liquidus Temperature (°C) | 1150 | – |
| Solidus Temperature (°C) | 1145 | – |
| Latent Heat of Crystallization (J/g) | 40 | – |
| Thermal Conductivity (W/cm·°C) | 0.05 | 0.0008 |
| Specific Heat Capacity (J/g·°C) | 0.25 | 0.45 |
| Density (g/cm³) | 7.2 | 1.6 |
| Interface | Thermal Resistance (cm²·°C·s/J) |
|---|---|
| Chill-Sand | 30 |
| Chill-Casting | 50 |
| Casting-Sand | 150 |
Analysis of Temperature Curves and Parameter Optimization
The experimental temperature curves exhibited a sharp rise within the first 5 minutes after pouring, followed by a slower ascent to peak temperatures. During cooling, the slope of the curves gradually decreased, indicating a reduction in cooling rate over time. When compared to initial CAE simulations, these curves showed substantial deviations, particularly in later stages where calculated temperatures fell significantly below measured values. This inaccuracy underscored the limitations of default parameters in ADSTEFAN for lost foam casting scenarios. To address this, we leveraged the software’s inverse calculation tool, inputting the actual temperature data from multiple points to refine the thermophysical properties. The optimization process involved systematic adjustments to thermal conductivity, specific heat, and latent heat values, resulting in a close alignment between simulated and experimental curves, as depicted in the comparative graphs.
Validation of the optimized parameters was conducted through a second experiment incorporating chill elements. The temperature profiles from four thermocouple points were compared against simulations using both initial and optimized parameters. The results demonstrated that optimized parameters reduced the standard deviation from 245 to 54, with simulated curves closely tracking experimental data across most intervals. Notably, temperature fluctuations observed in some measurements—attributed to sand cracking from thermal stress and gas pressure—were less pronounced in simulations, suggesting that the optimized models effectively captured average behavior. This consistency confirms the robustness of our approach for practical EPC applications, enabling more reliable predictions of thermal history and related phenomena.
Further analysis focused on the implications for defect prediction. Using the optimized parameters, we simulated shrinkage porosity based on the Niyama criterion and critical solid fraction flow. The results showed a marked improvement in accuracy, with predicted defect locations and severities aligning closely with actual observations in castings. For instance, in one凸模 casting, the optimized simulation correctly identified severe shrinkage in areas where minor defects were indicated by initial parameters. This enhancement underscores the importance of precise thermophysical data in CAE tools for lost foam casting, as it directly influences decisions on gating, chilling, and other process modifications to mitigate risks.
Investigating Cooling Rates and Microstructural Relationships
Beyond parameter optimization, we explored the correlation between cooling rates and microstructural evolution in ductile iron castings. Pearlite content, a key determinant of mechanical properties, is highly sensitive to cooling conditions. By fitting the temperature-time data from simulations to polynomial functions using numerical analysis software, we derived continuous curves that accurately represent the cooling behavior. The general form of these functions is expressed as:
$$ T(t) = a_0 + a_1 t + a_2 t^2 + \cdots + a_n t^n $$
where \( T \) is temperature, \( t \) is time, and \( a_0, a_1, \ldots, a_n \) are coefficients determined through regression. For each thermocouple point, the fitted curves achieved determination coefficients (R²) above 0.998, indicating excellent agreement with raw data. Differentiation of these functions yielded cooling rate profiles:
$$ \frac{dT}{dt} = a_1 + 2a_2 t + \cdots + n a_n t^{n-1} $$
These profiles revealed consistent trends across points within the same casting, emphasizing the uniformity of heat dissipation in lost foam casting processes. For example, points 1 and 2 from one casting displayed similar curve shapes, as did points 3 and 4 from another, despite variations in absolute values due to local geometry.
To quantify the impact on microstructure, we calculated the average cooling rate between 700 °C and 650 °C—the critical range for pearlite transformation—for each point. The results, combined with metallographic analysis of pearlite content, are summarized in Table 5. We observed a positive correlation between cooling rate and pearlite percentage, which we modeled using a quadratic equation:
$$ C_p = -42.3 \left( \frac{dT}{dt} \right)^2 + 11.8 \left( \frac{dT}{dt} \right) + 0.0331 $$
where \( C_p \) is the pearlite content and \( \frac{dT}{dt} \) is the cooling rate in °C/min. This relationship, though specific to our experimental conditions, provides a valuable tool for controlling microstructure in EPC processes. By manipulating cooling rates through strategic placement of chills or other methods, foundries can target desired pearlite levels in critical sections, enhancing product consistency and performance.
| Measurement Point | Average Cooling Rate (°C/min) | Pearlite Content (%) |
|---|---|---|
| 1 | 0.089 | 75.6 |
| 2 | 0.11 | 82.1 |
| 3 | 0.109 | 83.5 |
| 4 | 0.14 | 86.9 |
Conclusions and Industrial Implications
Our study demonstrates the efficacy of integrating CAE technology with temperature measurement tests to advance lost foam casting practices. By optimizing thermophysical parameters through inverse calculation, we achieved a significant improvement in simulation accuracy, with temperature predictions within 50 °C of experimental values. This precision extends to defect forecasting, enabling more reliable identification of shrinkage porosity and informing effective process adjustments. The optimized parameters, though tailored to our specific conditions, offer a replicable methodology for other foundries employing EPC techniques, particularly for thick-section ductile iron castings.
Moreover, the functional relationship between cooling rates and pearlite content provides a novel approach to microstructural control. Through data fitting and derivative analysis, we established a quantitative model that links thermal history to material properties, facilitating targeted interventions for enhancing pearlite formation. This insight is especially valuable in applications demanding high wear resistance or strength, where consistent microstructure is paramount. As lost foam casting continues to gain traction for complex, low-volume productions, our findings contribute to its refinement, reducing scrap rates and elevating quality standards. Future work could expand this framework to include additional alloy systems or advanced simulation features, further solidifying the role of CAE in modern foundry engineering.
In summary, the synergy between experimental data and computational tools unlocks new potentials for optimizing lost foam casting processes. Our research underscores the importance of continuous parameter validation and microstructural awareness, paving the way for more intelligent and efficient manufacturing in the casting industry.