In the context of nuclear power plant safety, the spray pump shell, as a critical nuclear-grade component, plays an indispensable role in ensuring operational integrity. The stringent service environment necessitates that these castings meet not only conventional technical specifications but also elevated requirements in mechanical properties, component stability, and durability. Consequently, the internal microstructure of the castings must be dense, free from defects such as shrinkage cavities, cracks, gas pores, and slag inclusions. During industrial production, after riser removal and rough machining, prominent casting defects like shrinkage cavities and porosity were observed on the upper surface and flange outer edge protrusions of the shell, severely compromising quality and hindering production efficiency. This study employs AnyCasting simulation software to analyze the casting process of the original design, identify the locations and causes of these defects, and implement process improvements that effectively eliminate shrinkage-related issues, ensuring casting quality and enhancing yield rates.
The spray pump shell casting, as illustrated in the provided model, features a complex geometry with significant wall thickness variations. Its maximum outer diameter is 755 mm, inner diameter is 315 mm, and height is 338 mm, with wall thicknesses ranging from 7 mm to approximately 90 mm. Weighing 350 kg, it falls under the category of medium-sized hub-like castings. The structure is generally symmetrical, with a bottom flange containing symmetric grooves and outer edge protrusions, as well as spiraling cavities around a central cylindrical through-hole. This complexity, coupled with substantial thickness differentials, presents considerable challenges in achieving sound castings.
The material specified is ZG0Cr13Ni4Mo, a low-carbon martensitic stainless steel known for its intricate phase transformations, including martensitic transformation and reversed austenite formation. When used for large or complex castings, improper processing can lead to deformation and cracking. Moreover, the alloy composition influences the liquidus temperature and solidification range, affecting the volume of liquid and solidification shrinkage. If this volumetric reduction is not compensated by feed metal, it results in shrinkage cavities and porosity. The casting characteristics of this material add another layer of difficulty to the production process.
To systematically address these casting defects, a detailed examination of the solidification behavior is essential. The formation of shrinkage cavities and porosity is fundamentally rooted in the volumetric changes during cooling. Metal contraction occurs in three stages: liquid contraction, solidification contraction, and solid contraction. The total volume change, $\Delta V_{total}$, can be expressed as:
$$ \Delta V_{total} = \Delta V_l + \Delta V_s + \Delta V_{so} $$
where $\Delta V_l$ is the liquid contraction, $\Delta V_s$ is the solidification contraction, and $\Delta V_{so}$ is the solid contraction. Liquid contraction depends on the temperature drop above the liquidus, often modeled as:
$$ \Delta V_l = \alpha_v \cdot \Delta T_l $$
Here, $\alpha_v$ is the volumetric thermal expansion coefficient of the liquid metal, and $\Delta T_l$ is the temperature decrease in the liquid state. Solidification contraction is related to the phase change and alloy composition, typically accounting for 3-6% for steels. If the sum of liquid and solidification contractions exceeds the external dimensional shrinkage and adequate feeding is absent, defects manifest. The feeding efficiency, $\eta_f$, can be defined as:
$$ \eta_f = \frac{V_{feed}}{V_{shrinkage}} \times 100\% $$
where $V_{feed}$ is the volume of metal supplied from risers and $V_{shrinkage}$ is the volumetric deficit due to contraction. Achieving $\eta_f \geq 100\%$ is critical to prevent casting defects.
The alloy composition of ZG0Cr13Ni4Mo significantly impacts its solidification behavior. The chemical composition is controlled within the ranges specified in Table 1. Elements like carbon, chromium, and nickel influence the solidification range and phase stability, directly affecting defect formation.
| Element | C | Si | Mn | S | P | Cr | Ni | Mo |
|---|---|---|---|---|---|---|---|---|
| Min | – | – | – | – | – | 12.0 | 3.5 | 0.4 |
| Max | 0.06 | 1.0 | 1.0 | 0.020 | 0.030 | 13.5 | 4.5 | 0.7 |
The original casting process utilized a bottom-gating system with an open pouring setup, employing refractory tubes for the sprue, runner, and ingates. Metal was introduced from the bottom of the mold via a single sprue and two symmetric runners. A layered risering approach was adopted, with an open top riser and a blind riser placed at lower levels, intended to provide sequential feeding. Chill blocks were arranged around the bottom flange face. Key process parameters for simulation are summarized in Table 2.
| Parameter | Value | Unit |
|---|---|---|
| Pouring Temperature | 1580 | °C |
| Pouring Time | 18 | s |
| Air-Metal Heat Transfer Coefficient | 41.87 | W/(m²·K) |
| Air-Mold Heat Transfer Coefficient | 41.87 | W/(m²·K) |
| Chill-Casting Heat Transfer Coefficient | 3000 | W/(m²·K) |
| Chill-Mold Heat Transfer Coefficient | 1000 | W/(m²·K) |
| Metal-Mold Heat Transfer Coefficient | Variable | W/(m²·K) |
Simulation of the original process using AnyCasting revealed three primary zones of casting defects. The first zone exhibited porosity on the upper flange surface, the second showed shrinkage cavities in the thicker inner wall section, and the third contained shrinkage cavities at the outer edge protrusions of the bottom flange. These results aligned with actual production issues. The defect formation is analyzed through thermal and solidification models. The temperature gradient, $G$, and solidification rate, $R$, are critical parameters. The condition for shrinkage cavity formation is often described by the $G/R$ ratio; low values promote dendritic growth and microporosity. The Niyama criterion, often used for steel castings, is expressed as:
$$ N_y = \frac{G}{\sqrt{\dot{T}}} $$
where $\dot{T}$ is the cooling rate. When $N_y$ falls below a threshold (e.g., 1 °C¹/²·s¹/²/mm for some steels), porosity is likely. In the problematic zones, localized hot spots and inadequate feeding paths led to low $G$ and high solidification times, triggering these casting defects.
The root causes of these casting defects are multifaceted. Firstly, alloy composition variations affect fluidity and shrinkage. The carbon content must be kept below 0.06% to maintain fluidity. The chromium-to-nickel equivalent ratio influences the solidification range; an imbalance can increase liquid shrinkage. Impurities like sulfur and phosphorus reduce fluidity and widen the freezing range, promoting defect formation. Secondly, the original process design failed to establish an optimal temperature gradient. The top riser, lacking insulation, cooled rapidly, reducing its feeding efficiency. The bottom-gating system created a reverse temperature gradient, with hotter metal at the bottom, leading to isolated hot spots in the mid-section. The thick protrusions at the flange edge acted as thermal centers, and the single blind riser provided insufficient feed metal, resulting in shrinkage cavities.
To address these issues, a comprehensive process improvement strategy was implemented. The modifications focused on enhancing feeding and controlling solidification patterns. The key changes are outlined in Table 3.
| Aspect | Original Process | Improved Process | Rationale |
|---|---|---|---|
| Top Riser | Open riser without insulation | Insulated exothermic riser sleeve | Prolongs solidification time, improves feeding efficiency |
| Additional Riser | None at central hole | Blind riser added at central through-hole | Directly feeds thick inner wall section |
| Chill Design | Chills only at bottom flange | Conformal chills added at outer edge protrusions | Accelerates cooling at hot spots, extends feeding range |
| Alloy Control | Standard composition | Stricter control of C, Cr/Ni ratio, and impurities | Optimizes fluidity and reduces shrinkage tendency |
The insulated exothermic riser ensures a longer liquid duration, enhancing feeding capability. The feeding capacity of a riser can be estimated by:
$$ V_{riser} \geq \frac{V_{casting} \cdot (\beta_l + \beta_s)}{\eta} $$
where $V_{casting}$ is the casting volume, $\beta_l$ and $\beta_s$ are the liquid and solidification contraction coefficients, and $\eta$ is the riser efficiency (typically 0.1-0.2 for conventional risers, but higher for exothermic types). For the spray pump shell, the total contraction volume was calculated based on empirical data. The liquid contraction coefficient for low-carbon stainless steel is approximately 1.5% per 100°C, and solidification contraction is about 4%. With a casting volume of approximately 0.045 m³, the required feed metal volume is:
$$ V_{shrinkage} = 0.045 \times (0.015 \times \Delta T_{liquid} + 0.04) $$
Assuming a $\Delta T_{liquid}$ of 150°C, $V_{shrinkage} \approx 0.0034$ m³. The improved riser system was designed to provide over 0.004 m³ of feed metal, ensuring adequate compensation.
The addition of conformal chills at the flange protrusions modifies the local cooling rate. The chill effect can be modeled using the heat transfer equation:
$$ q = h_c \cdot A \cdot (T_{casting} – T_{chill}) $$
where $q$ is the heat flux, $h_c$ is the heat transfer coefficient (set at 3000 W/(m²·K)), $A$ is the contact area, and $T$ denotes temperatures. By increasing $q$, the solidification time at the protrusion is reduced, shifting the thermal center toward the riser. This extends the effective feeding distance, $L_f$, which for steel castings can be approximated as:
$$ L_f = k \cdot \sqrt{T} $$
where $k$ is a material constant and $T$ is the section thickness. With chills, $L_f$ increases, allowing the riser to feed previously isolated regions.
Simulation of the improved process was conducted with the same parameters as before, but incorporating the modifications. The results demonstrated a significant reduction in casting defects. The solidification sequence showed a clear directional progression from the casting extremities toward the risers, with the risers being the last to solidify. The probability of defect formation, represented by a scalar field in AnyCasting, dropped to near-zero values in the casting body, with any residual porosity confined to the riser heads. The temperature distribution at critical time intervals is summarized in Table 4, highlighting the improved gradient.
| Location | Original Process Temperature at 50% Solidified (°C) | Improved Process Temperature at 50% Solidified (°C) | Change |
|---|---|---|---|
| Upper Flange Surface | 1420 | 1480 | +60 |
| Inner Wall Thick Section | 1450 | 1390 | -60 |
| Flange Outer Edge Protrusion | 1440 | 1350 | -90 | Top Riser Center | 1460 | 1520 | +60 |
The higher temperatures in the riser and upper flange indicate better feeding, while lower temperatures in the thick sections show reduced hot spots. The Niyama criterion values across the casting were computed post-simulation. For the improved process, over 95% of the casting volume exhibited $N_y > 2$ °C¹/²·s¹/²/mm, indicating a low risk of casting defects. In contrast, the original process had extensive zones with $N_y < 1$, correlating with defect locations.
The practical implementation of the improved process in production yielded castings that were fully dense, with smooth surfaces and no visible defects like cracks or sand inclusions. Non-destructive testing via ultrasonic inspection confirmed the absence of internal shrinkage cavities or porosity beyond acceptable limits. The yield rate increased substantially, addressing the production bottleneck. This success underscores the importance of integrating simulation tools like AnyCasting into the foundry workflow to predict and mitigate casting defects.
Further analysis of the solidification kinetics provides deeper insight. The solid fraction, $f_s$, as a function of time, $t$, can be derived from cooling curves. For a given volume element, the heat balance is:
$$ \rho C_p \frac{\partial T}{\partial t} = \nabla \cdot (k \nabla T) + \rho L \frac{\partial f_s}{\partial t} $$
where $\rho$ is density, $C_p$ is specific heat, $k$ is thermal conductivity, and $L$ is latent heat. Solving this equation numerically in AnyCasting allows tracking of $f_s$. In the original process, regions with slow $f_s$ evolution corresponded to defect zones. The improved process showed a more uniform $f_s$ progression, minimizing isolated liquid pools.
Another aspect considered was the effect of mold material. The use of resin sand, with its lower thermal conductivity compared to metallic molds, contributes to slower cooling. The thermal diffusivity, $\alpha$, given by:
$$ \alpha = \frac{k}{\rho C_p} $$
is lower for resin sand, prolonging solidification. This necessitates efficient risering. The exothermic riser compensates by providing additional heat, maintaining a favorable temperature gradient.
In conclusion, the study demonstrates that casting defects in complex components like the spray pump shell can be effectively analyzed and eliminated through numerical simulation and targeted process improvements. By adjusting riser design, adding chills, and controlling alloy composition, a sound casting was achieved. The methodology not only resolved the immediate production issue but also provides a framework for addressing similar casting defects in other applications. Future work could explore advanced feeding systems or real-time monitoring to further enhance quality and reduce reliance on post-casting inspections.