The presence of slag inclusions, also known as non-metallic inclusions, represents one of the most common and detrimental defect categories in metal castings. These defects typically originate from exogenous sources such as eroded refractory materials, dross, or fragmented oxide films entrained during melt transfer and mold filling. Slag inclusion defects act as discontinuities within the metallic matrix, creating severe stress concentration points that significantly degrade the mechanical properties of the cast component, particularly fatigue strength and fracture toughness. The morphology, size, distribution, and population density of these slag inclusions are critical factors influencing the initiation and propagation of fatigue cracks during service, ultimately determining the lifespan and reliability of the casting. Therefore, accurate prediction and effective control of slag inclusion defects are paramount concerns in the pursuit of high-integrity castings.
During mold filling and subsequent solidification, entrained slag particles are transported by the molten metal flow. Crucially, these particles do not remain isolated; they undergo complex interactions, including collision, adhesion, and aggregation. This dynamic aggregation-growth phenomenon leads to the formation of larger particle clusters. These clusters exhibit different thermo-physical behaviors compared to individual particles. Due to the non-wetting nature between slag and molten metal, aggregated clusters possess lower surface energy and are less likely to disperse. Furthermore, their increased volume and mass alter the balance of forces acting upon them—such as drag, buoyancy, and gravitational forces—which directly influences their trajectory and final resting position within the solidified casting. Predicting this behavior is highly complex, as it involves a stochastic process coupled with the transient thermal, fluid flow, and solidification conditions of the casting process.
Traditional numerical simulation approaches for tracking discrete particles often treat them as inert, non-interacting entities with constant size. This simplification limits the predictive accuracy for slag inclusion defects, as it neglects the dynamic changes in particle size distribution and the consequent effect on particle kinematics. To bridge this gap, this work presents the development and application of a Particle Aggregation-Growth Model (PAGM). This model, integrated within a coupled fluid flow and solidification simulation framework, aims to enhance the fidelity of slag inclusion prediction by accounting for the dynamic aggregation of slag particles during casting processes. The primary objective is to investigate the influence of different gating system designs on the final distribution and overall slag inclusion rate in a cast component, using a propeller casting as a case study.
1. The Particle Aggregation-Growth Model (PAGM)
The PAGM is built upon a Lagrangian particle-tracking framework within a continuum Eulerian description of the molten metal flow. The model incorporates physics-based criteria to govern the interaction and coalescence of discrete slag inclusion particles.
1.1 Fundamental Assumptions
To render the complex multi-physics problem computationally tractable, the following assumptions are made:
- All slag inclusion particles and their aggregates are treated as solid spheres.
- Upon aggregation, the total volume of the resulting cluster is the sum of the volumes of the constituent particles. The cluster maintains a spherical shape and is assumed to be irreversible (no break-up).
- The aggregation process follows a “larger particle absorbs smaller ones” rule within an interaction zone. The mass and volume of smaller particles are added to the larger host particle, and the smaller particles are subsequently removed from the computation.
- Considering the high shear forces in regions of high fluid velocity, particle aggregation is suppressed where the local fluid velocity exceeds a critical threshold.
- Momentum is conserved during the collision and aggregation event, with kinetic energy losses neglected.
1.2 Governing Equations for Particle Motion and Aggregation
The motion of a single slag particle in the molten metal is governed by Newton’s second law, considering dominant forces. The equation of motion in the k-direction (where k represents x, y, or z) is given by:
$$
\frac{\pi d_p^3 \rho_p}{6} \frac{d u_{p,k}}{dt} = (F_{p,k} + F_{g,k}) – (F_{d,k} + F_{b,k})
$$
Where:
$d_p$ is the particle diameter,
$\rho_p$ is the particle density,
$u_{p,k}$ is the particle velocity component,
$t$ is time,
$F_{p,k}$ is the force component due to the pressure gradient in the fluid,
$F_{g,k}$ is the gravitational force component,
$F_{d,k}$ is the drag force component exerted by the fluid, and
$F_{b,k}$ is the buoyancy force component.
The drag force is typically calculated using a correlation such as:
$$
F_{d,k} = \frac{1}{2} C_d \rho_f A_p |\vec{u_f} – \vec{u_p}| (u_{f,k} – u_{p,k})
$$
where $C_d$ is the drag coefficient, $\rho_f$ is the fluid density, $A_p$ is the particle’s projected area, and $\vec{u_f}$ is the local fluid velocity.
The core of the PAGM lies in the algorithm that evaluates potential aggregation events at each computational time step. For two particles, i and j, to be considered for aggregation, they must simultaneously satisfy three criteria:
1. Proximity Criterion: The distance between particle centers must be within a defined interaction zone, which scales with particle size.
$$
L \leq \frac{1}{2} L_{i\_lim} = \frac{1}{2} \kappa d_i
$$
Here, $L$ is the inter-particle distance, $L_{i\_lim}$ is the interaction range limit for particle i, $d_i$ is its diameter, and $\kappa$ is a model coefficient governing the interaction distance.
2. Absolute Velocity Criterion: Aggregation is disabled in high-velocity regions to account for shear-induced dispersion.
$$
|\vec{u_{p,i}}| \leq u_{lim}
$$
Where $u_{lim}$ is a user-defined velocity threshold.
3. Relative Velocity Criterion: The relative motion of the particles must indicate a converging trajectory. This is assessed by checking the sign of the relative velocity vector projected onto the vector connecting the two particles. Aggregation is activated if:
$$
(\vec{u_{p,i}} – \vec{u_{p,j}}) \cdot (\vec{x_i} – \vec{x_j}) < 0
$$
This condition implies the particles are moving towards each other.
When all criteria are met for a pair of particles, the larger particle (say, i) absorbs the smaller one (j). The properties of the host particle are updated as follows:
$$
d_{i,new} = \left( d_i^3 + d_j^3 \right)^{1/3}
$$
$$
\vec{u_{i,new}} = \frac{m_i \vec{u_{p,i}} + m_j \vec{u_{p,j}}}{m_i + m_j}
$$
where $m_i$ and $m_j$ are the particle masses. Particle j is then removed from the simulation.
2. Numerical Simulation of Slag Inclusion Distribution in a Propeller Casting
2.1 Casting Geometry and Simulation Setup
The test case is a marine propeller casting with a diameter of 140 mm, produced via sand casting using a ZL102 aluminum alloy. The key process parameters and PAGM settings for the baseline simulation are summarized in the table below.
| Parameter Category | Description | Value / Specification |
|---|---|---|
| Casting | Material | ZL102 Aluminum Alloy |
| Process | Pouring Temperature | 680°C |
| Pouring Rate | 0.3 kg/s | |
| Mold Type | Silica Sand | |
| Slag Particles (Simulated) | Material | SiO₂ |
| Mean Initial Diameter | 0.6 mm | |
| Density | 2.3 g/cm³ | |
| Total Mass Introduced | 1.25 g | |
| Approx. Number of Particles | 4,800 | |
| PAGM Parameters | Interaction Distance Coefficient ($\kappa$) | 1.5 |
| Velocity Threshold ($u_{lim}$) | 0.5 m/s |
The simulation couples the calculation of fluid flow (using the Navier-Stokes equations with a free surface model), heat transfer, and solidification. The discrete slag inclusion particles are introduced at the inlet and tracked using the PAGM equations throughout filling and solidification.
2.2 Analysis of Slag Particle Trajectory and Aggregation Behavior
The simulation reveals the dynamic and interconnected nature of flow and slag inclusion evolution. During the initial stage of filling, the metal-slag mixture enters the sprue. The high velocity in the vertical sprue inhibits particle aggregation due to the high shear environment. As the metal reaches the hub and diverges into the three blade passages, the flow velocity decreases significantly, and recirculation zones develop. In these regions, the proximity, velocity, and trajectory criteria are frequently satisfied, triggering the onset of slag inclusion aggregation.
Particles carried along the leading edge of the flow into the thin blade sections often encounter geometrical obstructions near the blade tip, leading to their capture. Meanwhile, in the blade root and hub regions, where multiple flow fronts meet and velocities are lowest, conditions are highly favorable for both the accumulation and aggregation of slag inclusion particles. Following mold filling, buoyancy forces become increasingly dominant. Larger aggregated clusters experience greater buoyant force and rapidly rise to the upper surfaces of the casting (top of the hub and upper blade surfaces). Smaller, non-aggregated particles may remain suspended within the bulk liquid and become trapped internally upon solidification.
The final predicted distribution shows a clear pattern: a high concentration of larger slag inclusion clusters on the upper (cope) surfaces of the blades and hub, with fewer and smaller inclusions in the lower sections and internal regions. This pattern is a direct consequence of the combined effects of flow-driven transport, dynamic aggregation which creates buoyant clusters, and gravity-driven flotation.
3. Influence of Gating System Design on Slag Inclusion Distribution
The design of the gating system is a critical factor controlling the flow characteristics and, consequently, the behavior of entrained slag. Four distinct gating system configurations were simulated and analyzed using the PAGM to quantify their effectiveness in minimizing the slag inclusion content in the final propeller casting.
- Top Gating: Metal enters directly at the top of the casting hub.
- Bottom Gating: Metal enters from beneath the casting through a simple runner bar.
- Bottom Gating with a Whirl Gate: A centrifugal trap (whirl gate) is incorporated into the runner system.
- Bottom Gating with a Top Riser: A bottom gate is used in conjunction with a large top riser placed on the hub.
The performance of each system was evaluated based on the simulated final distribution of slag inclusions and key metrics such as the total volume of slag trapped in the casting and the calculated “slag trapping efficiency” of the gating system itself. The trapping efficiency $p$ is defined as:
$$
p = \left( 1 – \frac{V_c}{V_0} \right) \times 100\%
$$
where $V_c$ is the total volume of slag inclusions found within the casting cavity proper, and $V_0$ is the total volume of slag particles introduced at the pour.
| Gating System Design | Maximum Slag Cluster Diameter (mm) | Total Slag Particles in System | Slag Particles in Casting | Slag Volume in Casting (cm³) | Gating System Slag Trapping Efficiency (p) |
|---|---|---|---|---|---|
| Top Gating | 2.73 | 998 | 299 | 0.131 | 74.74% |
| Bottom Gating | 2.30 | 1022 | 358 | 0.309 | 42.78% |
| Bottom Gating with Whirl Gate | 1.89 | 1246 | 236 | 0.198 | 63.33% |
| Bottom Gating with Top Riser | 2.27 | 1075 | 270 | 0.213 | 60.56% |
3.1 Comparative Analysis
The results demonstrate a significant influence of gating design on slag inclusion mitigation:
- Top Gating: This design yielded the highest slag trapping efficiency (~75%). The sprue itself acts as an effective slag trap. The turbulent entry promotes early aggregation in the sprue, creating large clusters that are easily separated from the metal flowing into the casting. Consequently, the casting contained the lowest volume of slag inclusion material. However, the potential for surface turbulence and oxide film entrainment with this method must be carefully considered.
- Bottom Gating: This system produced the worst outcome, with over 60% of the introduced slag volume ending up inside the casting. The smooth runner offers minimal resistance to slag particle entry, and the tranquil filling allows particles to be carried into the cavity where they eventually float up to the upper surfaces of the blades and hub.
- Bottom Gating with Whirl Gate: The addition of a centrifugal trap markedly improved performance. The whirl gate effectively captured a significant portion of the incoming slag particles by centrifugal force, preventing them from entering the casting. The trapping efficiency rose to ~63%. The slag clusters found in the casting were generally smaller, indicating that the most egregious aggregations were removed by the trap.
- Bottom Gating with Top Riser: This design also showed improvement over the simple bottom gate. The large top riser provides a reservoir of hot metal and a low-pressure region that encourages the flotation and collection of slag particles that have entered the mold cavity. It acts as a “final collector,” reducing the slag inclusion volume in the casting proper, resulting in a trapping efficiency of ~61%.
The simulation clearly illustrates the mechanisms: effective gating systems either prevent slag entry (like the whirl gate) or provide a designated, benign location for slag to collect (like the top riser or the sprue in top gating). The dynamic aggregation model is essential to this analysis, as the size evolution of particles directly impacts their kinematics and capture probability.
4. Conclusion
The formation and distribution of slag inclusion defects in castings are governed by a complex interplay of fluid dynamics, particle kinematics, and dynamic particle interaction phenomena. This study demonstrates that accounting for the aggregation and growth of slag particles is crucial for accurate defect prediction. The developed Particle Aggregation-Growth Model (PAGM) provides a more physically realistic framework for simulating slag inclusion behavior compared to traditional inert-particle tracking methods.
The application of the PAGM to an aluminum propeller casting revealed that the gating system design has a profound impact on the final slag inclusion distribution and the overall defect rate. Key findings include:
- Slag particles undergo significant dynamic aggregation during mold filling, particularly in low-velocity and recirculation zones. This aggregation increases cluster buoyancy, driving them towards the upper surfaces of the casting.
- Gating system design is a powerful tool for controlling slag inclusion defects. Top-gating systems can exhibit high intrinsic trapping efficiency by utilizing the sprue as a slag collection basin. For bottom-gating systems, incorporating features like centrifugal whirl gates is highly effective in preventing slag entry, while top risers are effective in collecting slag that has entered the mold cavity.
- Numerical simulation incorporating the PAGM can effectively predict the size and location of slag inclusion concentrations, offering valuable insights for optimizing gating and risering designs to minimize defect rates and improve casting quality and reliability.
This enhanced predictive capability supports more robust casting design and aids in failure analysis by pinpointing potential locations of critical defects stemming from slag inclusions. Future work may focus on refining the aggregation physics, including potential cluster breakup under shear, and validating the model across a wider range of alloy systems and casting geometries.