In the manufacturing of large-scale power generation equipment, components like the stationary blade holder ring are critical. This part, acting as a carrier for multiple stages of stationary blades within a steam turbine, directly influences the efficiency and operational stability of the entire unit. The manufacturing of such large, complex geometry components predominantly relies on sand casting processes. However, the inherent challenges of sand casting, particularly for parts with significant variations in wall thickness, often lead to the formation of various sand casting defects. These defects, primarily shrinkage porosity, shrinkage cavities, gas entrainment, and deformation, can severely compromise the mechanical performance and dimensional accuracy of the final casting, potentially leading to catastrophic failure in service.
Traditionally, optimizing the casting process to mitigate these sand casting defects relied heavily on empirical knowledge and costly trial-and-error methods. This approach is not only time-consuming but also inconsistent. With advancements in computational power, numerical simulation technology has emerged as a powerful tool for foundry engineers. By simulating the filling and solidification processes virtually, it allows for the prediction of potential sand casting defects before any metal is poured. This enables proactive optimization of process parameters, significantly reducing development costs and improving first-pass yield. In this study, I employed InteCAST (Huazhu CAE) simulation software, combined with the Design of Experiments (DOE) methodology, to systematically investigate the influence of key sand casting process parameters—pouring temperature, filling time, and mold initial temperature—on the formation of shrinkage and deformation defects in a ZG230-450 stationary blade holder ring produced via phenolic resin sand casting.
Materials, Methodology, and Experimental Design
The material studied was a carbon steel casting, ZG230-450. Its nominal chemical composition (in wt.%) is approximately 0.30% C, 1.20% Mn, 0.50% Si, with S and P contents kept below 0.040%. The mold was made from phenolic resin-bonded sand, a common choice for large steel castings due to its good collapsibility and strength.
The core of the methodology was the integration of numerical simulation and orthogonal experimental design. An L9(3^4) orthogonal array was chosen to efficiently study the three factors, each at three levels, requiring only nine simulation runs. The selected factors and their levels are summarized in Table 1.
| Factor | Level 1 | Level 2 | Level 3 |
|---|---|---|---|
| Filling Time (s) | 5 | 10 | 15 |
| Pouring Temperature (°C) | 1550 | 1570 | 1590 |
| Mold Initial Temperature (°C) | 10 | 20 | 30 |
The corresponding orthogonal design matrix, defining the nine distinct process combinations to be simulated, is presented in Table 2.
| Experiment No. | Filling Time (s) | Pouring Temperature (°C) | Mold Initial Temperature (°C) |
|---|---|---|---|
| 1 | 5 (Level 1) | 1550 (Level 1) | 10 (Level 1) |
| 2 | 5 (Level 1) | 1570 (Level 2) | 30 (Level 3) |
| 3 | 5 (Level 1) | 1590 (Level 3) | 20 (Level 2) |
| 4 | 10 (Level 2) | 1550 (Level 1) | 30 (Level 3) |
| 5 | 10 (Level 2) | 1570 (Level 2) | 20 (Level 2) |
| 6 | 10 (Level 2) | 1590 (Level 3) | 10 (Level 1) |
| 7 | 15 (Level 3) | 1550 (Level 1) | 20 (Level 2) |
| 8 | 15 (Level 3) | 1570 (Level 2) | 10 (Level 1) |
| 9 | 15 (Level 3) | 1590 (Level 3) | 30 (Level 3) |
Prior to simulation, the 3D CAD model of the casting, along with its gating and risering system, was discretized using a uniform mesh. The meshing scheme details are provided in Table 3. A uniform mesh, while computationally demanding for such a large part, provides consistent accuracy for predicting sand casting defects like shrinkage, which are highly sensitive to local solidification conditions.
| Parameter | Value |
|---|---|
| Total Number of Elements | 5.851 million |
| Number of Elements in Casting | 0.293 million |
| Element Edge Length | 18 mm |
| Total Molten Metal Mass | 19,777.29 kg |
| Yield | 65.72% |
Analysis of Simulation Results for Sand Casting Defects
The simulation of the solidification process for a representative case revealed the thermal history and progression of the solidus front. The process began with the loss of superheat from the thinner sections and areas in contact with chills. As solidification advanced, isolated liquid pockets formed in the middle-section diaphragm of the holder ring. These last-to-freeze zones, inadequately fed by the risers, are the prime locations for the formation of shrinkage-related sand casting defects. The total simulated solidification time was approximately 28,655 seconds.
Prediction and Quantification of Shrinkage Porosity and Cavities
The post-processing module of InteCAST was used to predict the location and volume of shrinkage porosity and cavities for all nine experimental runs. Figure 1 shows an illustrative example of the shrinkage porosity distribution predicted in the casting. The defects are predominantly concentrated in the middle diaphragm and, under certain conditions, in the upper crossbeam section.
Quantitative data for the volume of shrinkage porosity and cavities from each simulation are compiled in Table 4. Notably, significant shrinkage cavities were predicted only in Experiment No. 2, while shrinkage porosity was a persistent sand casting defect across most parameter combinations. The total volume of these shrinkage-related sand casting defects varied considerably, with Experiment No. 9 showing the minimum value.
| Exp. No. | Shrinkage Porosity Volume (cc) | Shrinkage Cavity Volume (cc) | Total Shrinkage Volume (cc) |
|---|---|---|---|
| 1 | 559.87 | 0.00 | 559.87 |
| 2 | 1702.94 | 81.65 | 1784.59 |
| 3 | 1084.75 | 0.00 | 1084.75 |
| 4 | 361.58 | 0.00 | 361.58 |
| 5 | 198.29 | 0.00 | 198.29 |
| 6 | 221.62 | 0.00 | 221.62 |
| 7 | 390.74 | 0.00 | 390.74 |
| 8 | 507.38 | 0.00 | 507.38 |
| 9 | 174.96 | 0.00 | 174.96 |
To determine the degree of influence of each process parameter on the total shrinkage volume, an analysis of means (ANOM) and analysis of range (Range Analysis) was performed. For each factor at a given level (i, ii, iii), the average total shrinkage volume was calculated from the three experiments containing that level. The range (R) for each factor is the difference between the maximum and minimum of these level averages. A larger range indicates a greater influence of that factor on the sand casting defects. The results are presented in Table 5.
| Factor / Level | Level 1 (K1, k1) | Level 2 (K2, k2) | Level 3 (K3, k3) | Range (R) |
|---|---|---|---|---|
| Filling Time (s) | K1=3429.21, k1=1143.07 | K2=781.49, k2=260.50 | K3=1073.08, k3=357.69 | 882.57 |
| Pouring Temp. (°C) | K1=1312.19, k1=437.40 | K2=2490.26, k2=830.09 | K3=1481.33, k3=493.78 | 392.69 |
| Mold Temp. (°C) | K1=1288.87, k1=429.62 | K2=1673.78, k2=557.93 | K3=2321.13, k3=773.71 | 344.09 |
The analysis reveals a crucial insight: Filling Time has the most significant influence (largest Range, R=882.57) on the total volume of shrinkage-related sand casting defects. The average shrinkage volume is lowest at a filling time of 10 seconds (k2=260.50 cc). Both shorter (5s) and longer (15s) filling times led to higher defect volumes. Pouring Temperature also shows a substantial effect. Interestingly, the intermediate pouring temperature of 1570°C resulted in the highest average shrinkage, while 1550°C gave the lowest. Mold Initial Temperature has a lesser, but still notable, influence, with a colder mold (10°C) being beneficial for reducing these sand casting defects under the tested conditions. The optimal combination from this analysis for minimizing shrinkage is: Filling Time: 10s, Pouring Temperature: 1550°C, Mold Initial Temperature: 10°C. However, this conflicts with the single best result from Experiment No. 9 (15s, 1590°C, 30°C), highlighting the complex interactions in sand casting processes and the need to consider other defects like deformation.
Analysis of Temperature Gradients and Prediction of Deformation
Deformation and hot tearing are other critical sand casting defects driven by thermal stresses during solidification. These stresses arise from non-uniform cooling and constraint, which are quantified by temperature gradients (G). I analyzed the evolution of temperature gradient at two critical locations: Node A in a thick section and Node B in a thin section (refer to original figure for location).
The temperature gradient at Node A (thick section) drops rapidly initially and then stabilizes. To compare the effect of process parameters, the gradient value at a fixed time of 1000 seconds was extracted for each run. The range analysis on this data (Table 6) shows that Pouring Temperature has the strongest influence on the mid-solidification gradient in thick sections. A higher pouring temperature generally leads to a lower temperature gradient at this stage, which can reduce thermal stress.
| Factor | k1 (°C/mm) | k2 (°C/mm) | k3 (°C/mm) | Range (R) |
|---|---|---|---|---|
| Filling Time | 93.58 | 90.08 | 93.03 | 3.50 |
| Pouring Temperature | 98.58 | 96.18 | 81.93 | 16.65 |
| Mold Temperature | 93.98 | 90.43 | 92.28 | 3.55 |
The temperature gradient at Node B (thin section) peaks around 530 seconds into solidification. Analyzing the peak gradient values (Table 7) confirms that Pouring Temperature is the dominant factor influencing the maximum thermal gradient in thin sections. A higher pouring temperature (1590°C) results in a higher peak gradient, which could increase the risk of deformation or cracking in constrained thin sections.
| Factor | k1 (°C/mm) | k2 (°C/mm) | k3 (°C/mm) | Range (R) |
|---|---|---|---|---|
| Filling Time | 5.17 | 5.18 | 5.22 | 0.05 |
| Pouring Temperature | 5.12 | 5.18 | 5.27 | 0.15 |
| Mold Temperature | 5.22 | 5.18 | 5.17 | 0.05 |
The relationship between pouring temperature (Tp), gradient (G), and the resulting strain (ε) that leads to deformation can be conceptualized. A simplified model for thermally induced strain is often expressed as:
$$ \epsilon = \alpha \cdot \Delta T + \frac{\sigma}{E} $$
Where $\alpha$ is the coefficient of thermal expansion, $\Delta T$ is the temperature difference (related to G), $\sigma$ is stress, and E is Young’s modulus. Higher gradients (G) lead to larger $\Delta T$ across small distances, increasing stress ($\sigma$) and consequently, the strain ($\epsilon$), potentiating sand casting defects like deformation. The simulation shows that a lower pouring temperature reduces the peak gradient in thin sections, likely reducing deformation risk, but our shrinkage analysis showed it might not be optimal for feeding. Conversely, the parameter set from Experiment No. 9 (high pouring temp) minimizes shrinkage but may increase thermal gradients. However, the actual deformation is a complex integral of stress over time and constraint. The best overall result likely balances these competing factors.
Discussion and Synthesis of Optimal Parameters
The investigation into sand casting defects for the stationary blade holder ring reveals a multi-objective optimization problem. The goal is to find a set of process parameters that simultaneously minimizes both shrinkage porosity and the risk of deformation.
From the analysis, two key insights emerge:
- Control of Shrinkage Defects: Filling time is the most critical parameter for controlling the volume of shrinkage-related sand casting defects. An intermediate filling time of 10 seconds provided the best average result in the orthogonal analysis. Extremely short times may cause turbulence and air entrainment, while long times increase heat loss and favor pasty solidification, worsening shrinkage. Pouring temperature also plays a vital role, but its effect is non-linear, requiring careful selection within a narrow window.
- Mitigation of Deformation Risk: Pouring temperature is the dominant factor influencing thermal gradients, which drive thermal stresses and deformation. A lower pouring temperature reduces peak gradients in thin sections, theoretically lowering deformation. However, too low a temperature can impair fluidity and feeding, exacerbating other sand casting defects.
The orthogonal analysis pointed to one optimal set for shrinkage (10s, 1550°C, 10°C), while the single best simulation run for low shrinkage volume was Experiment No. 9 (15s, 1590°C, 30°C). This discrepancy underscores the limitations of analyzing factors in isolation and the presence of strong interactions. Experiment No. 9, while having a higher pouring temperature, also has a longer filling time and a warmer mold. This combination likely creates a more favorable thermal history that promotes directional solidification towards the risers despite the higher temperature, effectively reducing isolated liquid pools and shrinkage. The warmer mold also slows the initial cooling rate, potentially reducing thermal gradients and stress, counteracting the effect of the high pouring temperature on deformation.
Therefore, based on the comprehensive simulation results—which showed Experiment No. 9 had the absolute lowest shrinkage volume and a manageable thermal profile—the recommended optimal process parameters to minimize overall sand casting defects for this specific stationary blade holder ring are:
- Filling Time: 15 seconds
- Pouring Temperature: 1590 °C
- Mold Initial Temperature: 30 °C
This parameter set represents a balanced solution that prioritizes soundness (minimizing shrinkage) while the thermal conditions suggest a controlled risk of deformation. Subsequent actual production trials using these optimized parameters confirmed the simulation predictions, yielding castings with higher density and less distortion compared to those produced with prior empirical methods, validating the effectiveness of the numerical simulation approach in controlling sand casting defects.
Conclusion
This study successfully demonstrates the application of numerical simulation and design of experiments to investigate and optimize the sand casting process for a large, complex steel casting. The formation of critical sand casting defects, namely shrinkage porosity and deformation, was systematically analyzed in relation to three key process parameters. The main conclusions are:
- Among the factors studied, filling time has the greatest influence on the total volume of shrinkage porosity and cavities. An appropriately controlled filling time is essential to promote favorable solidification patterns and minimize these sand casting defects.
- Pouring temperature is the most significant factor affecting the temperature gradients within the casting during solidification, which are the primary drivers for thermal stress and deformation-related sand casting defects.
- The optimal process parameters for the ZG230-450 stationary blade holder ring cast in phenolic resin sand were identified as a filling time of 15 seconds, a pouring temperature of 1590°C, and a mold initial temperature of 30°C. This combination provided the best compromise, yielding the minimum predicted shrinkage volume while maintaining a thermal history conducive to limiting deformation.
- The integration of numerical simulation provides a powerful, cost-effective methodology for predicting and mitigating sand casting defects prior to production, moving beyond reliance on traditional trial-and-error methods. The correlation between simulation predictions and improved production outcomes validates this approach as a critical tool for modern foundry engineering.
This work provides a theoretical and practical framework for optimizing sand casting processes for large, critical components, directly contributing to improved product quality and reliability by proactively addressing the root causes of sand casting defects.