In the production of sand castings for nuclear power plant components, internal discontinuities such as shrinkage porosity and macro-shrinkage cavities present significant challenges. My work focuses on a specific steam generator support casting made of ZG12MnMoV low-alloy steel. The original sand casting process produced unacceptable levels of sand casting defects, primarily shrinkage-related. By combining numerical simulation with systematic process redesign, I successfully eliminated these defects and improved the casting quality. This paper describes the defect analysis, the parametric optimization, and the verification through both simulation and production.
1. Introduction
The steam generator support is a critical safety-related component in pressurized water reactor nuclear power plants. It must withstand high temperatures, corrosive environments, and seismic loads over decades of service. The casting is produced by sand casting, a flexible but defect-prone process. Common sand casting defects in steel castings include shrinkage cavities, gas porosity, hot tears, and inclusions. Among these, shrinkage defects caused by improper directional solidification are the most detrimental. My investigation began with the observation that the original production run exhibited numerous internal shrinkage voids, leading to rejection during ultrasonic inspection. To address this, I employed the AnyCasting simulation software to model the filling and solidification behavior, identify the root causes, and devise an optimized mold design.
2. Casting Geometry and Material Characteristics
The steam generator support has a complex geometry with a maximum thickness of 200 mm, a minimum thickness of 70 mm, and an overall height of 1305 mm. The total weight is approximately 1.1 tons. The casting material, ZG12MnMoV, is a low-carbon micro-alloyed steel with the following nominal composition:
| Element | C | Si | Mn | S | P | Mo | V | Fe |
|---|---|---|---|---|---|---|---|---|
| Content | ≤0.15 | ≤0.6 | 1.2–1.7 | ≤0.02 | ≤0.025 | 0.2–0.4 | 0.05–0.1 | Balance |
This alloy undergoes peritectic solidification, which inherently increases the linear shrinkage rate and the tendency for micro‑porosity. The required mechanical properties after quenching and tempering include a tensile strength ≥500 MPa and a yield strength ≥400 MPa. The casting must also pass the MC2000 ultrasonic inspection standard, which demands fully dense internal soundness.
Common sand casting defects such as shrinkage, porosity, and hot tears are illustrated above.
3. Original Casting Process and Defect Analysis
The initial process used resin‑bonded sand molds, a three‑part flask, and a stepped gating system. A cylindrical riser (Φ420 mm × 540 mm) was placed at the top, and a side blind riser (Φ220 mm × 320 mm) was located at the mid‑wall transition. Several chills were applied near thick sections. The pouring temperature was 1580 °C.
I performed a solidification simulation using AnyCasting on the original design. The predicted shrinkage distribution showed a high volume fraction of porosity in the central thick region and near the junction between the main body and the side boss. The results matched the actual ultrasonic indications. The primary causes were identified as:
- Inadequate directional solidification: the long vertical distance prevented effective feeding from the top riser.
- Insufficient riser volume and height: the top riser solidified too early, losing its feeding capability.
- Incorrect placement of chills: some chills created isolated hot spots rather than promoting uniform temperature gradients.
The thermal gradient in the critical region can be described by the solidification time ratio between riser and casting. The modulus of the riser must satisfy:
$$ M_{\text{riser}} = k \cdot M_{\text{casting}} $$
where \(k\) is a safety factor typically between 1.2 and 1.5. In the original design, the top riser modulus was only 1.05 times the casting modulus, leading to premature freezing.
4. Process Optimization Measures
Based on the simulation insights, I redesigned the casting process with the following modifications:
4.1 Changed Parting Line and Pouring Position
I repositioned the main parting line to reduce the vertical height of the casting above the parting, allowing a single top riser to feed the entire geometry. This improved the feeding distance and reduced the number of risers required.
4.2 Riser and Feeder Design
I replaced the cylindrical top riser with a larger waist‑shaped insulated riser and added a second waist‑shaped blind riser at the opposite end. To extend the feeding range, I introduced a feeding aid (a metallic or insulating pad) along the thick walls. The thickness of the feeding aid was calculated using the corrected modulus method:
$$ \delta = \frac{E \cdot M_{\text{casting}}}{M_{\text{riser}} – M_{\text{casting}}} $$
where \(E\) is the ratio of thermal conductivities between the sand and the insulating pad material.
| Parameter | Value |
|---|---|
| Pouring temperature | 1550 °C |
| Pouring time | 34 s |
| Riser type (top) | Waist‑shaped insulated riser (480 mm × 350 mm × 600 mm) |
| Riser type (bottom) | Waist‑shaped blind riser (300 mm × 260 mm × 400 mm) |
| Feeding aid material | Insulating pad (thermal conductivity 0.5 W/m·K) |
| Chill material | Steel chills (thermal conductivity 45 W/m·K) |
4.3 Chill Placement
By analyzing the temperature field from the simulation, I placed chills at the thick‑thin transitions to create artificial chill zones that accelerate local solidification and prevent hot spots. The number and size of chills were optimized to maintain a progressive solidification front.
4.4 Pouring System Adjustment
I modified the gating system from a stepped to a bottom‑fill design to minimize turbulence and air entrapment, which also reduces the risk of oxide inclusions.
5. Numerical Simulation and Results
I used AnyCasting to simulate the optimized process. The pre‑processing included a non‑uniform mesh of 4,285,836 elements. The material properties and boundary conditions are listed below.
| Component | Value |
|---|---|
| Air | 41.87 |
| Steel chill | 3000 (constant high for effective chilling) |
| Insulating sleeve | 100 |
| Mold (furan resin sand) | Variable (temperature dependent) |
5.1 Filling Simulation
The filling simulation showed a smooth, non‑turbulent flow with a front velocity below 0.5 m/s, avoiding sand erosion and re‑oxidation. The entire cavity filled in 34 s as planned.
5.2 Solidification and Shrinkage Prediction
The solidification simulation predicted that the riser would remain liquid for at least 15 min longer than the thickest section of the casting, ensuring complete feeding. The predicted shrinkage porosity was completely eliminated in all critical zones. The temperature gradient along the casting length was calculated as:
$$ G = \frac{\Delta T}{\Delta x} $$
where \(\Delta T\) is the temperature difference between the riser and the farthest feeding point, and \(\Delta x\) is the feeding distance. The optimized design achieved \(G > 3 \text{ °C/cm}\), which is sufficient for sound solidification.
The solidification time at each location was extracted, and the Niyama criterion \(N = G / \sqrt{R}\) (where \(R\) is the cooling rate) was evaluated. All locations exhibited \(N > 1\) (mm·°C/s)^{1/2}, indicating freedom from micro‑shrinkage.
6. Production Verification
Based on the optimized simulation, I produced the steam generator support using the new mold design. After shakeout, the casting underwent visual inspection, dimensional checks, and ultrasonic testing per MC2000. The results were as follows:
- Surface: clean, no cracks, cold shuts, or sand inclusion.
- Internal quality: ultrasonic testing showed no indications exceeding the acceptance criteria.
- Mechanical properties: tensile strength 535 MPa, yield strength 425 MPa, elongation 22%.
- Yield: the casting yield improved from 65% to 78% due to reduced riser volume and elimination of defect‑related scrap.
The final casting is shown in the production photograph. The absence of sand casting defects confirmed the accuracy of the numerical simulation and the effectiveness of the process optimization.
7. Conclusions
Through this work, I demonstrated that the combination of AnyCasting numerical simulation and systematic process redesign can effectively eliminate sand casting defects in thick‑section steel castings. The key conclusions are:
- The original steam generator support casting suffered from shrinkage porosity due to improper riser placement, insufficient riser modulus, and non‑directional solidification.
- By modifying the parting line, adding a waist‑shaped insulated riser, introducing a feeding aid, and optimizing chill placement, the feeding distance was extended and a favorable thermal gradient was established.
- The simulation predicted zero shrinkage porosity after optimization, which was validated by actual production.
- The use of the corrected modulus formula and the Niyama criterion provided quantitative guidance for riser and chill design.
- This methodology is directly applicable to other large steel castings prone to sand casting defects, particularly in the nuclear and energy sectors.
Future work will focus on developing automated optimization loops that integrate AnyCasting with genetic algorithms to further reduce trial‑and‑error in process design.