I conducted an integrated numerical and experimental study on a top-loaded ball valve steel casting produced by sand gravity casting. The component is large, geometrically complex, and characterized by pronounced wall-thickness variation. In my work, the central technical problem is not merely filling the mold, but controlling the formation of casting defect populations such as shrinkage cavities, dispersed porosity, gas entrapment, and nonmetallic inclusions. I used ProCAST as the main simulation platform because its finite element formulation, coupled filling-solidification solver, and defect criteria can represent the thermal and fluid behavior of heavy steel castings with acceptable engineering accuracy. The optimized route I developed combines gating system design, riser-chill coordination, mesh and boundary-condition control, orthogonal experiment design, and production validation. The final result is a reproducible process that moves the main casting defect risk into the riser, improves the qualified product rate, shortens the production cycle, and reduces energy consumption.
Engineering Context and Casting Defect Risk
The casting I studied is a top-loaded ball valve body made of WCB cast steel. Its mass is approximately 5.76 t, and its overall envelope is about 1920 mm × 1406 mm × 1310 mm. This size class means that the solidification time is long, the thermal gradient is difficult to control, and the feeding path must remain open for a long period. The top-loaded design offers maintenance and sealing advantages in service, but from a foundry perspective it creates a large cavity, multiple thick sections, and several hot spots. These features make the component highly sensitive to casting defect formation. In particular, shrinkage porosity and shrinkage cavities are expected when the liquid metal cannot compensate for liquid and solidification contraction. Gas porosity and inclusions are also possible when the filling front is unstable or when the gating system fails to trap slag and air.
I organized the baseline process around sand gravity casting because the component is large, the batch size is significant, and resin-bonded sand provides sufficient strength, collapsibility, and surface quality. The initial gating concept was bottom gating, with a vertical sprue, radial runners, and a symmetric arrangement of in-gates. This choice was made to reduce turbulence and to promote a stable rising front. However, bottom gating also creates a temperature field that can be unfavorable for feeding, because the lower part of the mold may cool too quickly while the upper part remains hot. Therefore, I had to coordinate the gating system with risers and chills to suppress casting defect formation.
| Parameter | Baseline Value |
|---|---|
| Casting type | Top-loaded ball valve body |
| Material | WCB cast steel |
| Mass | Approximately 5.76 t |
| Overall size | 1920 mm × 1406 mm × 1310 mm |
| Casting method | Sand gravity casting |
| Mold material | Resin-bonded sand |
| Gating type | Bottom gating |
| Initial pouring temperature | 1570 C |
| Initial mold temperature | 20 C |
| Pouring rate | 100 kg/s |
Material Behavior and Thermal Properties
Because ProCAST does not initially contain every specific steel grade, I defined WCB as a user material and calculated its thermophysical response from composition. The liquidus and solidus temperatures were approximately 1509 C and 1472 C, respectively. These values control the freezing range and therefore the tendency for mushy-zone casting defect formation. A wide freezing range usually increases the risk of dispersed microporosity, while a narrow freezing range can produce concentrated shrinkage cavities. For WCB in this heavy section, both types of casting defect were possible, so I used Niyama criterion maps together with solid-fraction and temperature-field analysis.
| Element | Specified Range (wt.%) | Typical Allowance |
|---|---|---|
| C | 0.25-0.30 | ±0.01 |
| Si | ≤0.60 | ±0.05 |
| Mn | ≤1.00 | ±0.05 |
| P | ≤0.040 | ±0.005 |
| S | ≤0.045 | ±0.005 |
| Cr | ≤0.50 | ±0.20 |
| Ni + Mo | ≤1.00 | ±0.02 |
| Property | Requirement | Typical Range |
|---|---|---|
| Tensile strength, Rm | ≥485 MPa | 500-620 MPa |
| Yield strength, Rp0.2 | ≥250 MPa | 270-350 MPa |
| Elongation, A | ≥22% | 25-32% |
| Hardness | 137-179 HB | 150-180 HB |
Gating System Design and Filling Control
I compared top gating, bottom gating, middle gating, and step gating before selecting bottom gating for this casting. The comparison is not only hydraulic; it also affects the thermal field and the final casting defect distribution. Top gating gives a favorable downward temperature gradient for feeding, but it produces severe splashing and air entrainment in a large cavity. Middle gating is a compromise but is difficult to place in this geometry. Step gating can be effective in tall thin-wall components, but it introduces complex flow control and possible “random pouring” behavior. For the top-loaded ball valve body, I selected bottom gating because it provides a calm rising front, reduces sand erosion, and minimizes gas entrapment. The penalty is a reversed thermal gradient, which I compensated for with risers and chills.
| Gating Type | Filling Stability | Slag Trapping | Feeding Gradient | Typical Casting Defect Risk |
|---|---|---|---|---|
| Top gating | Low | Moderate | Favorable | Gas porosity and inclusion defects |
| Bottom gating | High | High | Unfavorable unless aided | Shrinkage defect in upper hot spots |
| Middle gating | Moderate | Moderate | Moderate | Local shrinkage and cold shut |
| Step gating | Moderate to high | Moderate | Favorable | Complex flow-related casting defect |
The filling time must match the casting mass, gating capacity, and allowable temperature loss. I used the following relation to estimate the required filling duration:
$$ t_{\text{fill}} = \frac{M_{\text{poured}} f_{\text{fill}}}{100 \dot m} $$
where \(M_{\text{poured}}\) is the total poured mass including the casting, risers, and gating system; \(f_{\text{fill}}\) is the target fill percentage; and \(\dot m\) is the mass flow rate. For the baseline process, the total poured mass was approximately 9000 kg, the target fill was 100%, and the mass flow rate was 100 kg/s, giving a fill time near 90 s. In the optimized process, the filling time increased slightly to about 105 s because the flow resistance and thermal mass changed. This slower filling was intentional: it reduced the risk of gas entrainment and improved the stability of the rising front, which directly reduced the probability of gas-related casting defect formation.
Riser Design and Feeding Physics
Riser design is the main defense against shrinkage-related casting defect formation. I used the modulus method as the first design tool. The modulus \(M\) is the ratio of volume to cooling surface area:
$$ M = \frac{V}{S} $$
For a sound casting, the riser modulus \(M_r\) must be greater than the modulus of the feeding region \(M_c\). In practice I used:
$$ M_r \ge 1.1 M_c $$
For side blind risers and riser necks, the relationship must also satisfy the feeding path requirement. I used the following typical ratios:
$$ M_c : M_n : M_r = 1 : 1.1 : 1.2 $$
where \(M_n\) is the modulus of the riser neck. The solidification time follows the Chvorinov relationship:
$$ t_s = k M^2 = k \left(\frac{V}{S}\right)^2 $$
This equation explains why a larger modulus delays solidification. My original design placed four腰-shaped? I must avoid Chinese characters. I will write “waist-shaped” or “oblong” risers at the flange-main-body connection and cylindrical risers at other hot spots. The intended function was to keep the riser liquid longer than the casting hot spot. However, the original simulation showed that the riser top solidified too early, which closed the feeding channel and allowed shrinkage casting defect formation in the casting body.
To correct this, I replaced the ordinary sand risers with exothermic-insulating risers. The main top risers received a 20 mm thick FT400 exothermic sleeve. The middle auxiliary risers were changed to a double-layer design, and insulating cover plates were added on the top. The effective modulus of an insulating riser can be represented as:
$$ M_{\text{eff}} = \frac{V}{a S_{\text{side}} + b S_{\text{top}}} $$
where \(a\) and \(b\) are reduction factors for the side and top heat loss. For an insulating blind riser, \(a\) and \(b\) are typically about 0.7. This increases the effective modulus and extends the time during which the riser can feed the casting. In my simulation, the exothermic sleeve increased the useful feeding time and moved the final casting defect location into the riser. The feeder efficiency can be estimated by:
$$ \eta = \frac{V_c + V_r}{V_r} \times 100\% $$
where \(V_c\) is the volume contraction of the casting section being fed and \(V_r\) is the initial riser volume. By using exothermic and insulating materials, I increased the effective feeding efficiency and reduced the amount of liquid steel that had to be sacrificed in the gating system.
| Riser Type | Feeding Efficiency Range | Best Use in This Casting |
|---|---|---|
| Plain sand riser | 12-15% | Noncritical small hot spots |
| Waist-shaped sand riser | 12-15% | Flange hot spots in baseline design |
| Exothermic-insulating riser | 25-30% | Main top risers after optimization |
| Atmospheric riser | 15-20% | Auxiliary feeding when geometry allows |
| Insulating cover with sleeve | 30-60% in combined practice | Heavy sections and final feeding zone |
Chill Design and Directional Solidification
Chills are essential when the natural thermal gradient does not favor feeding. In this casting, the lower flange and thick central regions required additional cooling so that the solidification front would advance toward the risers. I estimated the chill thickness using:
$$ t_{\text{chill}} = K \delta $$
where \(K\) is a material coefficient and \(\delta\) is the equivalent hot-spot thickness. For steel castings, \(K\) usually lies between 0.3 and 0.8. I used this relation to adjust the chill thickness after the first simulation. I also used the working surface area relation:
$$ S_y = \frac{V_0 (M_0 – M_1)}{M_0 M_1 (1-y)} $$
where \(S_y\) is the required chill working area, \(V_0\) is the casting volume, \(M_0\) is the original modulus, \(M_1\) is the target modulus after chilling, and \(y\) is the chill expansion coefficient. This formula allowed me to place chills where they reduced the local modulus enough to eliminate dispersed porosity without creating cold shuts or hard spots.
| Interface | Heat Transfer Coefficient, W/m²·K |
|---|---|
| Casting-mold | 300-1000 |
| Casting-air | 5-10 |
| Mold-air | 5-10 |
| Casting-chill | 1000-5000 |
| Chill-mold | 300-1000 |
Numerical Model and ProCAST Setup
I built the casting, gating system, risers, chills, and virtual sand box in a 3D CAD environment and imported the assembly into ProCAST through the Visual-Mesh module. I used nonuniform mesh sizes to balance accuracy and computation time. The sand box was meshed at 100 mm, the large risers at 50 mm, the smaller risers at 30 mm, and the casting and chills at 10 mm. The final model contained approximately 90,000 surface elements and 460,000 volume elements. This mesh density was sufficient to resolve the filling front and the thermal gradients in the thick sections without making the simulation impractical.
| Region | Mesh Size (mm) | Reason |
|---|---|---|
| Sand box | 100 | Reduce total element count |
| Large risers | 50 | Resolve feeding behavior |
| Auxiliary risers | 30 | Capture neck solidification |
| Casting body | 10 | Resolve hot spots and defects |
| Chills | 10 | Capture steep thermal gradients |
The boundary conditions I used follow the standard heat-transfer and fluid-flow framework. For the mold-casting interface, I applied a coincident boundary condition to allow a temperature discontinuity. For the casting and the gating system, I used an equivalent boundary condition because the material is continuous. For the chill-casting interface, I applied a high heat-transfer coefficient. The main thermal boundary conditions are:
$$ q = h (T_f – T_w) $$
$$ q = \varepsilon \sigma_0 T_s^4 $$
where \(h\) is the convective heat-transfer coefficient, \(T_f\) is the fluid temperature, \(T_w\) is the wall temperature, \(\varepsilon\) is the emissivity, \(\sigma_0\) is the Stefan-Boltzmann constant, and \(T_s\) is the absolute surface temperature. The energy equation I solved in the solidification stage is:
$$ \rho c_p \left( \frac{\partial T}{\partial t} + u \frac{\partial T}{\partial x} + v \frac{\partial T}{\partial y} + w \frac{\partial T}{\partial z} \right) = \lambda \left( \frac{\partial^2 T}{\partial x^2} + \frac{\partial^2 T}{\partial y^2} + \frac{\partial^2 T}{\partial z^2} \right) + Q $$
The latent heat release during solidification was modeled through the enthalpy method:
$$ H = \int_0^T C_p \, dT + L f_s $$
where \(H\) is enthalpy, \(C_p\) is specific heat, \(L\) is latent heat, and \(f_s\) is solid fraction. The filling flow was solved using the mass, momentum, and volume-of-fluid equations:
$$ \frac{\partial \rho}{\partial t} + \nabla \cdot (\rho \mathbf{V}) = 0 $$
$$ \rho \left( \frac{\partial u}{\partial t} + u \frac{\partial u}{\partial x} + v \frac{\partial u}{\partial y} + w \frac{\partial u}{\partial z} \right) = -\frac{\partial P}{\partial x} + \mu \nabla^2 u + \rho g_x $$
$$ \rho \left( \frac{\partial v}{\partial t} + u \frac{\partial v}{\partial x} + v \frac{\partial v}{\partial y} + w \frac{\partial v}{\partial z} \right) = -\frac{\partial P}{\partial y} + \mu \nabla^2 v + \rho g_y $$
$$ \rho \left( \frac{\partial w}{\partial t} + u \frac{\partial w}{\partial x} + v \frac{\partial w}{\partial y} + w \frac{\partial w}{\partial z} \right) = -\frac{\partial P}{\partial z} + \mu \nabla^2 w + \rho g_z $$
For defect prediction, I relied mainly on the Niyama criterion. The criterion is defined by the ratio of the local temperature gradient \(G\) to the square root of the cooling rate \(R\):
$$ \frac{G}{\sqrt{R}} < C_{\text{Niyama}} $$
The local gradient was calculated from neighboring element temperatures:
$$ G = \max \left( \frac{T_{i+1,j,k} – T_{i,j,k}}{\sqrt{\Delta x^2 + \Delta y^2 + \Delta z^2}} \right) $$
The cooling rate was calculated from the temperature change over a time step:
$$ R = \frac{T_{i,j,k}^{m} – T_{i,j,k}^{m+1}}{\Delta t} $$
When the Niyama value falls below a critical threshold, the interdendritic feeding resistance becomes too high, and a casting defect in the form of microporosity is likely. For large steel castings, I used a critical value near 1.1. This criterion is not a substitute for physical understanding, but it is a powerful way to rank process designs before production.
| Simulation Parameter | Value |
|---|---|
| Filling template | Gravity filling |
| Termination temperature | 1462 C |
| Maximum time steps | 50,000 |
| Temperature output frequency | Every 10 steps |
| Initial pouring temperature | 1570 C |
| Initial mold temperature | 20 C |
| Mass flow rate | 100 kg/s |
| Fill limit | 100% |
| Stress model | Linear elastic for alloy |
Baseline Simulation and Casting Defect Analysis
The baseline simulation predicted smooth filling with a rising metal front. The filling process was stable, and no severe cold shut was observed. However, the solidification stage revealed a serious thermal problem. The riser top began to solidify before the casting hot spots had finished feeding. In the solid-fraction maps, the lower and middle regions reached high solid fraction while the upper flange-main-body connection remained liquid. This means the feeding path was reversed: the riser was no longer the last region to freeze. As a result, the casting could not be fed effectively, and shrinkage-related casting defect formation was predicted.
The baseline defect prediction showed four concentrated shrinkage cavities and several dispersed porosity zones. The main shrinkage cavities were located near the flange-main-body transition and around some thick-wall connections. The dispersed porosity was concentrated in the lower thick sections and in regions where the thermal gradient was low. The predicted total defective volume was about 13% of the casting volume in the Niyama map. This does not mean that 13% of the casting was useless, but it indicates that a significant fraction of the casting volume was at risk. Production inspection confirmed the simulation: ultrasonic testing and magnetic particle testing found shrinkage indications in the same regions. This agreement gave me confidence that the ProCAST model was capturing the dominant physics of casting defect formation.
| Baseline Result | Observation |
|---|---|
| Filling behavior | Stable bottom-gated rising front |
| Fill time | Approximately 90 s |
| Solidification sequence | Riser top solidified too early |
| Thermal gradient | Unfavorable for upward feeding |
| Predicted macro shrinkage | Four major locations |
| Predicted microporosity | Lower thick sections and flange transitions |
| Predicted defective volume | About 13% in Niyama map |
| Actual NDT result | Shrinkage indications consistent with simulation |
I also monitored temperature and solid fraction at several nodes. The nodes in the thick upper sections cooled more slowly than the nodes in the lower and central sections. More importantly, some lower nodes finished solidification before the upper thick section had entered the final freezing stage. This is exactly the condition that produces isolated liquid pockets and feeds the formation of shrinkage casting defect populations. The baseline process therefore failed not because the gating system was unstable, but because the feeding system could not maintain a continuous liquid channel to the last-solidifying regions.
Optimized Process Design
Based on the baseline simulation and production inspection, I made three major changes. First, I replaced the ordinary top risers with exothermic-insulating risers. The main top risers received a 20 mm FT400 exothermic sleeve. The middle auxiliary risers were converted to a double-layer insulating design, and insulating cover plates were placed on top. Second, I added two groups of chills in the flange region. Third, I adjusted the chill size and position on the pattern to improve local cooling without creating cold shuts. The objective was to create a controlled sequence: the thin and lower sections freeze first, the thick sections freeze later, and the riser remains liquid until the end. This is the classic condition for eliminating shrinkage casting defect formation.
| Optimization Measure | Purpose | Expected Effect on Casting Defect |
|---|---|---|
| Exothermic sleeve on main risers | Delay riser solidification | Improves feeding, moves shrinkage into riser |
| Double-layer auxiliary risers | Increase effective modulus | Reduces microporosity in transition zones |
| Insulating cover plates | Reduce top heat loss | Maintains liquid feeding channel |
| Additional flange chills | Accelerate local cooling | Reduces dispersed porosity in thick sections |
| Adjusted chill size | Control cooling rate | Avoids cold shut and hard spots |
The optimized simulation showed a much better solidification sequence. At the end of filling, the thin sections and chill regions had already begun to solidify, while the thick hot spots remained liquid. As solidification progressed, the freezing front advanced from the bottom and outer surfaces toward the top risers. The last liquid regions were located inside the insulating risers, not in the casting body. This means the risers successfully sacrificed themselves to feed the casting. The Niyama defect map after optimization showed that almost all predicted shrinkage was confined to the riser and the riser platform. The casting body showed no significant macro shrinkage and only minor microporosity risk at noncritical locations.
| Comparison Item | Baseline | Optimized Geometry |
|---|---|---|
| Main risers | Plain sand | Exothermic-insulating with 20 mm sleeve |
| Auxiliary risers | Single-layer sand | Double-layer insulating design |
| Top cover | None | Insulating cover plate |
| Flange chills | Limited | Two additional chill groups |
| Solidification end | Riser platform and casting transition | Inside riser |
| Predicted macro shrinkage | Four major locations | Confined to riser |
| Predicted microporosity | Lower thick sections | Greatly reduced |
| Risk of casting defect in final part | High | Low |
Orthogonal Experiment for Parameter Optimization
After improving the feeding geometry, I still needed to optimize the process parameters. I selected three factors: pouring temperature, pouring rate, and mold initial temperature. The response variable was the predicted porosity percentage from the Niyama criterion. I used an L9 orthogonal array with three factors and three levels. This design allowed me to evaluate the main effects with only nine simulations instead of twenty-seven full-factorial runs. I chose this approach because it is efficient, statistically valid, and practical for foundry process development.
| Level | Factor A: Pouring Temperature (C) | Factor B: Pouring Rate (kg/s) | Factor C: Mold Temperature (C) |
|---|---|---|---|
| 1 | 1580 | 100 | 20 |
| 2 | 1570 | 90 | 25 |
| 3 | 1560 | 105 | 30 |
| Run | A | B | C | Porosity (%) |
|---|---|---|---|---|
| L1 | 1580 | 100 | 20 | 12.35 |
| L2 | 1580 | 90 | 25 | 11.37 |
| L3 | 1580 | 105 | 30 | 13.62 |
| L4 | 1570 | 90 | 30 | 10.38 |
| L5 | 1570 | 105 | 20 | 7.54 |
| L6 | 1570 | 100 | 25 | 9.99 |
| L7 | 1560 | 105 | 25 | 14.64 |
| L8 | 1560 | 100 | 30 | 11.65 |
| L9 | 1560 | 90 | 20 | 15.53 |
The range analysis is summarized below. \(K_i\) is the sum of the response values at level \(i\), \(k_i\) is the average response at level \(i\), and \(R\) is the range. A larger \(R\) means that the factor has a stronger effect on the porosity-based casting defect risk.
| Range Item | Factor A | Factor B | Factor C |
|---|---|---|---|
| K1 | 37.52 | 37.37 | 33.99 |
| K2 | 27.91 | 30.56 | 37.28 |
| K3 | 41.82 | 39.14 | 35.80 |
| k1 | 12.50 | 12.46 | 11.33 |
| k2 | 9.30 | 10.19 | 12.43 |
| k3 | 13.94 | 13.05 | 11.93 |
| R | 13.91 | 8.58 | 3.29 |
The range analysis shows that the influence order is:
$$ \text{Pouring temperature} > \text{Pouring rate} > \text{Mold temperature} $$
For minimizing the porosity-related casting defect risk, the best levels are A2, B1, and C1. Therefore, the optimized combination is:
$$ T_{\text{pour}} = 1570 \text{ C}, \quad \dot m = 100 \text{ kg/s}, \quad T_{\text{mold}} = 20 \text{ C} $$
I then ran a full ProCAST simulation with this parameter set and the optimized riser-chill geometry. The predicted porosity decreased significantly compared with both the baseline and the earlier geometry-only optimization. The final predicted porosity was approximately 6.9%, and the remaining shrinkage indications were located in the upper part of the risers. This is an acceptable result because the riser is removed after casting. The casting body itself showed no significant shrinkage-related casting defect risk in the Niyama map.
Simulation Validation and Production Verification
To validate the optimized process, I followed the same production route used for the baseline casting: pattern preparation, resin-bonded sand molding, core assembly, melting in an induction furnace, refining and degassing, controlled pouring, slow cooling, shakeout, riser cutting, grinding, heat treatment, and nondestructive testing. I paid particular attention to chemical composition before pouring because small variations in carbon, silicon, and manganese can shift the freezing range and change the tendency for casting defect formation. After casting, I visually inspected the riser cross-section and the casting surface. The riser showed a sound feeding behavior: the lower part of the riser was dense, and the shrinkage cavity was located in the upper riser region. This is exactly what a well-designed riser should do.
I then carried out visual testing, magnetic particle testing, ultrasonic testing, and penetrant testing on the cleaned and ground casting. The results showed no macroscopic casting defect in the casting body. Minor indications were confined to the riser or to removable process material. The actual casting met the specification, and the result agreed with the ProCAST prediction. This agreement is important because it shows that the model can be used as a decision tool for heavy steel castings, not just as a visualization aid.
| Verification Step | Result |
|---|---|
| Chemical composition check | Within WCB specification |
| Pouring temperature | 1570 C |
| Pouring rate | 100 kg/s |
| Mold temperature | 20 C |
| Filling behavior | Stable and complete |
| Riser feeding | Shrinkage located in upper riser |
| Casting body NDT | No macroscopic casting defect found |
| Specification compliance | Passed |
Economic and Sustainability Benefits
The optimized process improved the qualified product rate from 78% to 92%. The production cycle decreased from 18 days to 13.5 days, and energy consumption decreased by approximately 15%. These improvements come directly from reducing the number of rejected castings, shortening the trial-and-error loop, and avoiding unnecessary remelting. In my analysis, the economic benefit has both direct and indirect components. Direct benefits include fewer scrapped castings, higher productivity, and lower energy cost. Indirect benefits include quality premiums, shorter delivery times, and improved customer retention.
| Economic Item | Baseline | Optimized | Improvement |
|---|---|---|---|
| Qualified product rate | 78% | 92% | +14 percentage points |
| Production cycle | 18 days | 13.5 days | -25% |
| Energy consumption | Baseline | Reduced by 15% | -15% |
| Annual scrap reduction | 110 pieces | 40 pieces | 70 fewer pieces |
| Annual direct benefit | Baseline | 24.425 million CNY | Significant gain |
| Annual indirect benefit | Baseline | 11.20 million CNY | Market and quality gain |
The annual direct benefit was estimated from reduced scrap, increased output, and energy savings. The indirect benefit was estimated from quality premium and market expansion. For a single heavy casting, the savings are already large; when the same methodology is applied to other pump and valve castings, the cumulative effect becomes much greater. More importantly, the optimized process reduces the chance that a casting defect reaches the final machined component, which improves safety and reliability in severe service conditions.
Extension to Other Castings
I also applied the same “simulation-analysis-optimization-validation” workflow to other heavy castings, including a shell-type component. I used the same mesh strategy, boundary-condition setup, and Niyama-based casting defect prediction. The results again showed that shrinkage indications were concentrated in the riser rather than in the functional body. This confirmed that the workflow is not limited to one geometry. The main requirements are accurate material data, a realistic heat-transfer coefficient, a well-designed gating system, and a clear understanding of the feeding path. When these conditions are met, ProCAST can guide process changes before the first casting is poured.
Key Lessons I Learned
First, a stable filling front does not guarantee a sound casting. In this component, the baseline filling simulation looked acceptable, but the solidification simulation revealed a fatal thermal imbalance. The most dangerous casting defect was not caused by turbulence but by the loss of the feeding channel. Second, riser geometry alone is not enough. The thermal properties of the riser sleeve and cover are just as important as the riser size. Third, chills must be placed with a clear purpose. Adding chills everywhere can create cold shuts, while adding too few chills leaves dispersed porosity in the thick sections. Fourth, orthogonal experiment design is a practical way to optimize parameters without exhausting the computational budget. Fifth, production validation is essential. The simulation is a map, but the foundry process is the territory.
Final Conclusions
I designed and optimized a sand gravity casting process for a top-loaded ball valve steel body. The baseline process used bottom gating, ordinary sand risers, and limited chills. The baseline simulation predicted four major shrinkage cavities and several microporosity zones, with a total defect-risk volume of about 13%. Actual nondestructive testing confirmed these shrinkage-related casting defect indications. I then redesigned the feeding system using exothermic-insulating risers, double-layer auxiliary risers, insulating top covers, and additional flange chills. The optimized solidification sequence moved the last liquid region into the riser and dramatically reduced the casting defect risk in the casting body.
I further optimized the process parameters using an L9 orthogonal experiment with pouring temperature, pouring rate, and mold temperature as factors. The best combination was 1570 C pouring temperature, 100 kg/s pouring rate, and 20 C mold temperature. The validation simulation showed that the remaining shrinkage was confined to the upper riser. Production verification showed no macroscopic casting defect in the casting body, and the casting met the required specification. The qualified product rate increased from 78% to 92%, the production cycle decreased by 25%, and energy consumption decreased by 15%. These results demonstrate that a ProCAST-based process optimization route can be used effectively for heavy steel castings with complex geometry and high casting defect sensitivity.
For future work, I recommend extending the material database, coupling macroscopic solidification with microscopic grain-structure prediction, and applying the same workflow to other pump and valve castings. If the material data, boundary conditions, and production feedback are continuously improved, the numerical model can become an even more reliable tool for reducing casting defect formation, shortening development time, and supporting sustainable foundry production.