Chapter 1. Introduction and Research Background
1.1 Background and Significance of the Research
The nuclear main pump is the sole rotating component in the reactor coolant system of a pressurized water reactor (PWR) nuclear power plant. It directly determines not only the reliability of the reactor type but also significantly influences the key economic parameters of the plant. The bearing support studied in this research is a critical load-bearing component of the reactor coolant pump in a PWR nuclear power station. The working environment inside the nuclear island is extremely harsh: high temperature, high pressure, high humidity, and intense radiation. Equipment replacement due to frequent malfunctions is practically impossible. Therefore, the sealing, durability, and reliability requirements for the pump assembly—and consequently for each of its components—are exceptionally stringent.
Nuclear energy ranks second only to thermal power in global energy utilization due to its high efficiency and cleanliness. As technology advances, an increasing number of countries are paying close attention to developing mature third-generation nuclear power technologies. In China, the CPR1000 (China Pressurized Reactor) program was developed based on the French M310 reactor design, incorporating improvements from advanced nuclear power technology concepts. This indigenous design, which is safe, efficient, and mature, has guided the construction of a fleet of “Generation II+” nuclear power stations.
In the PWR primary loop, the coolant water absorbs the enormous heat released by nuclear fission in the reactor pressure vessel. It then flows through the steam generator, where the heat is transferred to the secondary loop to generate steam and drive the turbine generator. The main pump provides the driving force to circulate the coolant water through this closed loop. Since the coolant water must remain liquid at temperatures of 300–350°C, the entire system operates under elevated pressure. This harsh combination of pressure, temperature, and radiation places extremely demanding requirements on the manufacturing quality of all pump components, particularly castings like the bearing support that must maintain structural integrity over prolonged service lifetimes.
1.2 Research Status of Microalloyed Cast Steels
The bearing support material is designated as 12MDV6 by AFNOR (the French standard organization). Its Chinese equivalent is ZG12MnMoV, which belongs to the family of microalloyed cast steels. In these steels, the mass fraction of each microalloying element is generally less than 0.10%. The most commonly used microalloyed cast steel systems include vanadium-based, niobium-based, and boron-based compositions.
Boron-based microalloyed cast steels significantly enhance hardenability. They offer good economic efficiency due to the rich boron resources in China. The boron content must be carefully controlled; excessive boron increases brittleness. During melting, oxygen and nitrogen levels must be controlled, typically by deoxidizing with aluminum and titanium before adding boron.
Vanadium and niobium-based microalloyed cast steels originated in Europe and exhibit excellent strength, toughness, weldability, and mechanical properties. With very low carbon and sulfur contents, these steels possess exceptional toughness and weldability. The vanadium and niobium ultimately form Nb(C,N) and V(C,N) precipitates, which refine the grain structure. Molybdenum aids in grain refinement and effectively retards the precipitation of niobium carbonitrides in austenite, ultimately strengthening the cast steel.
Rare-earth microalloyed cast steels are widely used in China due to the abundant rare-earth resources. Rare-earth elements powerfully purify the molten steel, improve the morphology and distribution of non-metallic inclusions, and enhance the steel’s microstructure.
The research status of microalloyed cast steels is summarized in the following table.
| Region | Institution/Researcher | Key Contribution |
|---|---|---|
| China | Kang Yonglin, Mao Xinping | Using Ti microalloying in low-carbon steel to raise yield strength to 450–700 MPa for weather-resistant steel |
| China | Northeastern University | Development of 700–780 MPa tensile strength beam steel using Nb-Ti microalloying |
| China | Lu Jiangxin et al. | Development of steel with yield strength exceeding 700 MPa by adding Ti, Nb, Mo |
| China | CISRI & Wuhan Iron and Steel | High-strength corrosion-resistant Nb-containing steel via deformation-induced ferrite transformation |
| USA | Bethlehem Steel (Burns Harbor) | Production of high-performance microalloyed steel plates for naval applications under the Defense Production Act |
| Japan/Korea | 1997 national projects | 10-year programs to develop next-generation steel materials with doubled strength |
| Japan | JFE (Funakawa, Shiozaki) | Ti-Mo microalloyed steel with tensile strength >780 MPa and excellent hole expansion ratio |
| Europe | Reip, Shanmugam | Ti-containing low-carbon steel for pipelines with tensile strength >600 MPa |
| Japan | Kamibayashi, Tanabe | Demonstrated that Ti addition is more effective than Nb for improving mechanical properties of low-carbon steel |
The development trends for microalloyed cast steels include: a renewed understanding of the beneficial effects of nitrogen and titanium; applications in building construction for improved fire resistance; the pursuit of ultra-fine grain technologies to concurrently enhance strength and toughness; and efforts to utilize harmful elements like sulfur and phosphorus in environmentally friendly ways.
1.3 Development of Casting CAE Technology
Casting process CAD and numerical simulation technology have become increasingly important in modern foundry practice. These tools enable engineers to numerically simulate the filling and solidification of castings, predict and analyze defects, and optimize processes without the need for expensive and time-consuming physical trial runs. This dramatically shortens development cycles, lowers costs, and improves quality.
The origins of casting simulation trace back to the 1960s, when researchers first applied finite difference methods to heat transfer calculations during solidification. Since then, numerous commercial software packages have emerged. Internationally, these include SIMULOR and Procast from France, CastCAE from Finland, ForCAST from Spain, and MAGMA from Germany. In China, the Huazhu CAE software developed by Huazhong University of Science and Technology and the Casting Star software developed by Tsinghua University have been widely adopted by domestic foundries. These tools can simulate temperature fields, flow fields, and even microstructural evolution, allowing accurate prediction of defects such as shrinkage cavity, porosity, sand burn-on, cold shuts, and inclusions.
“Huazhu CAE” is a sophisticated simulation system independently developed by Huazhong University of Science and Technology. After many years of refinement, it has formed a mature casting process simulation and analysis system. It can accurately model temperature fields and flow fields during casting. By analyzing the filling and solidification processes, it predicts defect locations, thereby avoiding defects and obtaining quality-satisfying castings. It supports various common casting materials; if the material is not in the built-in property database, users can edit and add it. Huazhu CAE is applicable to a wide range of casting methods and alloys. It not only analyzes solidification cooling, filling, and heat exchange coupling processes but also predicts defects including shrinkage cavity, shrinkage porosity, sand burn-on, and cold shuts. In this research, I used the Huazhu CAE software for macroscopic analysis of the filling and solidification processes, and used Procast software for preliminary microscopic simulation of the microstructure.
1.4 Main Research Contents
The specific objectives of this research are outlined below:
- Simulate the original foundry process of the main pump bearing support using Huazhu CAE, analyzing temperature fields and flow fields to predict defect locations.
- Identify the root causes of the shrinkage defects in the casting.
- Improve the casting process design and verify via simulation and production. Develop a solution for the sand burn-on (sand foundry defects) problem at hot spots by optimizing the coating process.
- Use uniform design and Minitab regression analysis to optimize the pouring process parameters.
- Investigate the influence of pouring temperature on the microstructure, particularly the secondary dendrite arm spacing, using Procast software.
Chapter 2. Casting Geometry and Original Production Process
2.1 Casting Features
The nuclear main pump bearing support is illustrated schematically in Figure 2.1 of the original thesis. Its overall dimensions are 1000 mm × 1000 mm × 1103 mm. The wall thickness exceeds 200 mm and is relatively uniform. The casting mass is approximately 2.5 tons. The material is ZG12MnMoV steel. Due to the large wall thickness, defects such as shrinkage porosity, shrinkage cavity, and sand burn-on are extremely likely to occur, posing a significant challenge for the foundry process design.
Using the hot spot circle calculation method, a substantial hot spot was identified at the intersection of the triangular rib and the vertical face of the bearing support. This region, marked as location A in the original thesis, is thick and prone to shrinkage defects unless properly fed. If the feeding system is inadequate, this hot spot will produce shrinkage porosity and cavity, rendering the casting unsuitable for service.
The chemical composition and mechanical properties of ZG12MnMoV are presented in the following tables.
| Element | C | Si | Mn | P | S | Mo | V |
|---|---|---|---|---|---|---|---|
| Content | ≤0.15 | ≤0.60 | 1.20–1.70 | ≤0.025 | ≤0.020 | 0.20–0.40 | 0.05–0.10 |
| Property | Yield Strength σs (MPa) | Tensile Strength σb (MPa) | Elongation δ (%) |
|---|---|---|---|
| Value | ≥400 | ≥500 | ≥18 |
ZG12MnMoV is a microalloyed cast steel located in the peritectic region of the Fe-C phase diagram. During the peritectic reaction, linear contraction occurs, which can lead to micro-shrinkage porosity. Additionally, manganese promotes grain coarsening. Therefore, a well-designed process is essential to obtain a dense, defect-free casting. The stringent service requirements mandate inspection to MC2000 ultrasonic testing standards to ensure the absence of internal shrinkage cavity and porosity defects.
2.2 Principles of Casting Process Design
Several fundamental design principles guided the process development for this casting.
Pouring position. For thick-section castings with significant volume contraction, directional solidification is preferred. The thickest sections should be positioned in the upper part of the mold to facilitate feeding from risers. Large flat faces should be placed at the bottom of the mold to avoid sand defects and to benefit from gravity-assisted feeding.
Parting line. The parting line should preferably place the entire casting, or at least its major part, in a single flask to ensure dimensional accuracy and minimize mismatch. The main machining datum planes should be in the same flask. The number of parting lines and cores should be minimized.
Gating system. A well-designed gating system must deliver metal smoothly and continuously into the mold cavity, avoiding turbulence that can cause gas entrapment and inclusions, control flow velocity and direction, fill the cavity within a reasonable time, and effectively retain slag.
Riser system. Sufficiently large modulus risers must ensure that solidification in the riser lags behind the regions they feed. The feeding path must remain open until solidification is complete. Adequate feeding pressure must overcome flow resistance.
Chills. Externally applied chills increase the solidification rate locally, improve the feeding distance of risers, control the solidification sequence, and eliminate hot spots. Chills must be clean, smooth, and properly positioned.
2.3 Original Process Design and Parameters
The original process used a step-gate gating system, sand casting with self-hardening resin sand for both the mold and cores, and manual molding. The coating was a water-based zircon flour refractory coating applied by brushing. Given the casting size, a flask of 1500 mm × 1500 mm was used. The casting received a quench-and-temper heat treatment.
Pouring position and parting line. The casting was oriented with its base downward. The parting line was located as shown in the original thesis. The intent was to place the large flat base at the bottom of the mold for soundness, but the feeding path to the hot spots at the triangular ribs was inadequate.
Gating system. The original gating system (Figure 2.3 in the thesis) was a step-gating system constructed with refractory brick tubes. The key dimensions were:
| Component | Dimension (mm) |
|---|---|
| Sprue | φ80 |
| Runner | φ80 |
| Ingate | φ50 |
| Ladle nozzle | φ45 |
| Pouring cup orifice | φ80 |
Riser system. An insulating blind riser (Riser No. 1) of waist-round shape was placed at the circular top of the bearing support, with dimensions a = 200 mm, b = h = 300 mm, and a sleeve thickness of 30 mm. Two circular insulating blind risers (Riser No. 2, φ200 mm, sleeve thickness 15 mm) were placed on the two triangular ribs. This design proved inadequate for feeding the massive hot spots.
Chills. The original process used two bar-shaped chills (Chill No. 1, 700 × 50 × 50 mm) on the top of the triangular ribs, and eleven rectangular chills (Chill No. 2, 200 × 160 × 80 mm) on the sides of the triangular ribs and the base of the circular support.
Chapter 3. Numerical Simulation and Analysis of the Original Process
3.1 Three-dimensional Modeling and STL File Import
I created the three-dimensional models of the main pump bearing support casting, gating system, risers, and chills using UG NX software. The models were exported as STL files and imported into the Huazhu CAE pre-processing module. The complete original process solid model is shown in Figure 3.1 of the thesis.
3.2 Pre-processing Stage
The pre-processing module performs mesh generation, material type selection, material priority selection, and mesh parameter settings. I selected the priority order as: casting, core, chill, riser sleeve, and mold, with priority decreasing from casting to mold.
Since the casting has a relatively uniform, simple geometry, uniform mesh division was chosen. For the pure solidification calculation, I selected a grid size of 7.7 mm, yielding 8,172,840 mesh cells. For the coupled flow-solidification calculation, the mesh was coarsened to approximately 600,000 cells to improve computation speed while retaining essential accuracy.
| Simulation Mode | Mesh Size (mm) | Number of Cells |
|---|---|---|
| Pure solidification | 7.7 | 8,172,840 |
| Coupled flow-solidification | ~16 | ~600,000 |
3.3 Calculation Stage
Huazhu CAE offers four calculation modes: pure solidification, pure flow, coupled flow-solidification, and coupled solidification + flow. For solidification analysis, I chose the pure solidification mode. For filling analysis, I used both pure flow and coupled calculations. Gravity feeding was activated to accurately predict shrinkage defects. The pouring time was set at 90 seconds. The computation ended when the casting solidification ratio reached 96%. The total number of saves was 100.
The material property parameters for ZG12MnMoV are listed below.
| Parameter | Value |
|---|---|
| C content (wt%) | 0.120 |
| Si content (wt%) | 0.350 |
| Mn content (wt%) | 1.360 |
| S content (wt%) | 0.020 |
| P content (wt%) | 0.011 |
| Mo content (wt%) | 0.280 |
| V content (wt%) | 0.070 |
| Liquidus temperature (°C) | 1511 |
| Solidus temperature (°C) | 1469 |
| Pouring time (s) | 90 |
3.4 Post-processing and Results Analysis
3.4.1 Filling Process Simulation and Analysis
The temperature field at different filling stages was examined. At 15–50% filling, the lowest temperature occurred at the two sides of the base away from the ingates (1515°C), attributed to the chilling effect of the base chills. The base center remained at higher temperature due to its large section, indicating a hot spot requiring careful feeding. At 87–96% filling, the lowest temperature (1512°C) was found at the triangular ribs, confirming the chills were functioning. The highest temperatures were at the risers on the triangular ribs, indicating the insulating sleeves were working.
The flow field analysis revealed that molten metal entered the base through the ingates and filled the base slowly. The flow velocity was relatively high at the ingate–base junction and at the junction between the triangular ribs and the riser. Higher velocity increases the risk of mold erosion and gas entrapment, both contributing to sand foundry defects and inclusions.
3.4.2 Solidification Process Simulation and Analysis
The original process adopted a layered feeding scheme intended to use the risers to feed the hot spots that solidify first, sequestering defects within the risers. However, the simulation results (Figure 3.8 in the thesis) revealed that while the top riser fed the vertical arc section effectively, the circular blind risers on the triangular ribs had insufficient modulus. Their feeding capacity was inadequate for the massive hot spot at the junction of the triangular rib and the vertical face. The simulation clearly showed a large amount of shrinkage cavity in the triangular rib region. Subsequent ultrasonic inspection of the trial-produced casting confirmed extensive internal shrinkage defects, rendering the casting unusable.
Chapter 4. Casting Process Improvement and Simulation
4.1 Process Analysis and Specific Improvements
The root cause of the sand foundry defects was identified as the insufficient modulus of the circular blind risers placed on the triangular ribs. Given the constraints of the original pouring position, these risers could not adequately feed the hot spot at the junction of the triangular rib and the vertical face. My improvements addressed this by fundamentally changing the casting orientation and redesigning the risering system.
Pouring position and parting line. I inverted the casting orientation so that the large base face was positioned at the top of the mold. This rearranged the solidification sequence to provide excellent accessibility for placing large, effective risers directly above the hot spot regions. The parting line was adjusted accordingly. This orientation permitted the use of open risers that were previously impossible to place.
Riser system redesign. The original insulating blind risers were replaced with a large insulating open riser placed on the base of the casting. This design was driven by the need to
- Provide a large enough modulus to outlast the solidification of the massive hot spots;
- Supply sufficient liquid metal volume to fully compensate for solidification shrinkage;
- Allow easy placement without the constraints of the triangular rib geometry;
- Simplify subsequent riser removal operations.
The new riser dimensions were φ560 mm × 900 mm with a sleeve thickness of 112 mm. The much larger riser modulus, combined with the favorable top-position feeding, provided substantially improved feeding efficiency compared to the original inadequate blind risers.
Chill design redesign. The original long strip chill on the triangular rib exceeded 200 mm in length and tended to deform after repeated use, compromising dimensional accuracy and weakening the chilling effect. I replaced it with 24 cylindrical chills of φ30 mm × 70 mm placed along the fillet at the junction of the triangular rib and the vertical face. Additionally, 10 bar-shaped chills of 180 mm × 120 mm × 90 mm were placed on the four circular bosses and the top arc section of the casting. This enhanced local cooling promoted directional solidification and increased the effective feeding distance of the top riser.
4.2 Simulation and Analysis of the Improved Process
4.2.1 Pre-processing
The priority order was unchanged: casting, core, chill, riser sleeve, mold. For pure solidification, the mesh size was 7.8 mm with 8,250,468 cells. For coupled calculations, the cell count was again approximately 600,000.
| Simulation Mode | Mesh Size (mm) | Number of Cells |
|---|---|---|
| Pure solidification | 7.8 | 8,250,468 |
| Coupled flow-solidification | ~16 | ~600,000 |
4.2.2 Post-processing and Results Analysis
Temperature field. During filling, the lowest temperatures were observed at the bottom arc region due to chills, with a minimum of 1493°C at the bosses. The temperature gradient was well-controlled, preventing cold shuts while maintaining a smooth thermal progression. The maximum temperature at the end of filling remained close to the pouring temperature in the riser, confirming good insulation.
Flow field. The improved process produced a slower, more controlled filling sequence. The molten metal first filled the arc region, then rose through the vertical walls, filled the base, and finally entered the top riser. The flow velocities were significantly lower than the original, minimizing mold erosion and reducing the risk of entrainment defects. This is critical because turbulent flow can create oxide films and gas bubbles, which are among the primary mechanisms that cause sand foundry defects in thick-section castings.
Solidification process. The pure solidification simulation result showed a clear progression of solidification toward the riser. The riser was the last region to solidify, as confirmed by the sequence of solidification fronts. The final image displayed no shrinkage cavity or porosity within the casting itself, indicating excellent feeding. The improved process successfully achieved directional solidification and complete feeding of the critical hot spots.
4.3 Analysis of Sand Burn-on Defect and Coating Process Improvement
4.3.1 The Function of Foundry Coatings
Foundry coatings serve several essential purposes: they smooth the casting surface and reduce surface roughness; they prevent sand adhesion and burn-on; they shield the casting from decomposition products of resin-bonded sand; and they facilitate easy shakeout and cleaning.
4.3.2 Relationship between Sand Burn-on and Coatings
Sand burn-on defects (a category of sand foundry defects) are classified into mechanical and chemical types. Mechanical burn-on occurs when liquid metal penetrates the pores and gaps on the mold surface and envelops sand grains, forming a metal-ceramic layer upon solidification. In thick-section castings, the high ferrostatic pressure and prolonged solidification time exacerbate this penetration. Chemical burn-on results from physico-chemical reactions between the metal, mold material, and mold gases, producing complex low-melting-point silicates that firmly bond sand to the casting. Coatings prevent these defects by sealing the surface pores, providing a thermal barrier, and, in some cases, generating a reducing atmosphere that aids in stripping the sintered layer.
4.3.3 Composition of Foundry Coatings
Foundry coatings generally consist of five components: refractory filler, carrier liquid, binder, suspending agent, and additives. The refractory filler, being the most critical, determines the coating’s anti-burn-on performance. Important filler properties include refractoriness, thermal conductivity, particle size, chemical inertness at high temperatures, sintering behavior, and strippability.
Water-based and alcohol-based coatings are the two main categories by carrier. The choice depends on the process, casting type, and specific conditions. The original process used a water-based zircon flour coating with the following composition:
| Component | Zircon Flour | Bentonite | Sugar Syrup | Coal Tar |
|---|---|---|---|---|
| Mass ratio | 100 | 2 | 3 | 1.5 |
4.3.4 Coating Application Methods
Common application methods include dipping, brushing, flow coating, spraying, and transfer coating. Dipping is suitable for small- to medium-sized cores. Brushing is simple, versatile, and well-suited for large or complex molds in single-piece or small-batch production. Flow coating is efficient and materialsaving but requires careful control. Spraying offers high productivity and uniform coatings, especially for large surfaces.
4.3.5 Analysis of Sand Burn-on Defect and Coating Improvement
In the original trial production, severe sand burn-on was observed at the hot spot region where the triangular rib meets the vertical face, while other regions were clean. The mechanism was clear: at the massive hot spot, the metal remained above 1540°C for a prolonged period, exceeding the decomposition temperature of zircon flour. Under sustained high temperature, zircon (ZrSiO₄) decomposes into SiO₂ and ZrO₂. The steel reacts with air to form FeO, which combines with SiO₂ to form low-melting fayalite (Fe₂SiO₄). This slag continuously erodes and melts the coating, eventually consuming the refractory filler entirely. Once the coating is breached, liquid metal penetrates directly into the sand mold, creating a thick, tenacious penetration layer. At other locations, the faster cooling rate preserved the zircon coating, which formed a dense sintered protective layer that prevented burn-on.
To resolve this at the troublesome hot spot while maintaining economic efficiency, I developed the following approach:
- Employ chromite sand (FeO·Cr₂O₃) as the facing sand exclusively at the hot spot region on the mold cavity surface corresponding to the triangular rib junction. Chromite sand has an extremely high refractoriness exceeding 1900°C and excellent thermal conductivity. At high temperature, it sinters rapidly, closing the intergranular voids and forming a dense barrier that prevents metal penetration.
- Use chromite sand with a particle size distribution of 40–70 mesh, applied at a thickness of 40–50 mm.
- Use furan resin sand as the backing sand.
- Brush the entire mold cavity surface with zircon flour coating.
- Replace the sugar syrup binder with water-soluble phenolic resin (for ambient temperature) and silica sol (for high-temperature bonding). Sugar syrup loses its binding effect at high temperature and is prone to fermentation in summer.
- Increase the number of brush coats from one to five, achieving a total coating thickness of 1.2–1.5 mm, sufficient to withstand the prolonged thermal exposure in a thick-section casting.
This refinement successfully eliminated sand burn-on defects at the hot spot region. The improved coating system allowed the entire casting to be cleaned easily and the final product passed the MC2000 ultrasonic inspection. Although the unit cost of chromite sand is higher, the elimination of defect repair and manual cleaning outweighed the cost, making the approach economically viable.
Chapter 5. Process Parameter Optimization Based on Uniform Design
5.1 Principles of Uniform Design
Uniform design is an experimental optimization method that seeks to represent the whole experimental domain with a minimal number of trials. In contrast to orthogonal design, which requires “uniform dispersal and comparable alignment” (often requiring at least n² trials for n levels), uniform design requires only n trials for n levels, making it highly efficient for multi-factor, multi-level optimization problems. The selection of test points follows the theory of uniform distribution in number theory, ensuring that test points are uniformly scattered throughout the experimental domain.
For a design with m factors and n levels, the number of experiments required is:
$$N_{\text{uniform}} = n$$
whereas for the same problem:
$$N_{\text{orthogonal}} = n^{2}$$
$$N_{\text{full}} = n^{m}$$
The uniform design table is denoted as Un(qs) or U*n(qs), where n is the number of trials, q is the number of levels per factor, and s is the maximum number of columns (factors) that can be accommodated.
5.2 Selection of the Uniform Design Table
For this study, I identified three factors: A — pouring temperature (°C); B — pouring time (s); and C — initial mold temperature (°C). Each factor was assigned five levels, as shown in the table below. Since the casting process inevitably involves variability, the objective was to identify the combination that minimizes total shrinkage volume and Niyama porosity, thereby minimizing sand foundry defects and internal discontinuities.
| Level | A: Pouring Temperature (°C) | B: Pouring Time (s) | C: Mold Initial Temperature (°C) |
|---|---|---|---|
| 1 | 1550 | 80 | 10 |
| 2 | 1555 | 90 | 20 |
| 3 | 1560 | 100 | 30 |
| 4 | 1565 | 110 | 40 |
| 5 | 1570 | 120 | 50 |
The uniform design table U*10(108) was selected. According to its usage table, for three factors, columns 1, 5, and 6 should be used. This yields the following 10 experiments:
| Trial No. | A (°C) | B (s) | C (°C) |
|---|---|---|---|
| 1 | 1550 | 120 | 20 |
| 2 | 1555 | 120 | 30 |
| 3 | 1560 | 110 | 50 |
| 4 | 1565 | 110 | 10 |
| 5 | 1570 | 100 | 20 |
| 6 | 1550 | 100 | 40 |
| 7 | 1555 | 90 | 50 |
| 8 | 1560 | 90 | 10 |
| 9 | 1565 | 80 | 30 |
| 10 | 1570 | 80 | 40 |
5.3 Experimental Results
Each of the 10 experiments was simulated using Huazhu CAE with the specified parameters. Two output responses were recorded: (1) the total shrinkage cavity volume obtained from the shrinkage porosity module; and (2) the total Niyama shrinkage porosity volume obtained from the Niyama criterion module.
| Trial No. | A (°C) | B (s) | C (°C) | Shrinkage Volume (cc) | Niyama Porosity Volume (cc) |
|---|---|---|---|---|---|
| 1 | 1550 | 120 | 20 | 25,680 | 4,450 |
| 2 | 1555 | 120 | 30 | 25,940 | 4,523 |
| 3 | 1560 | 110 | 50 | 26,370 | 4,432 |
| 4 | 1565 | 110 | 10 | 26,640 | 5,482 |
| 5 | 1570 | 100 | 20 | 26,980 | 5,931 |
| 6 | 1550 | 100 | 40 | 25,960 | 7,157 |
| 7 | 1555 | 90 | 50 | 26,430 | 6,956 |
| 8 | 1560 | 90 | 10 | 26,620 | 6,742 |
| 9 | 1565 | 80 | 30 | 27,140 | 5,920 |
| 10 | 1570 | 80 | 40 | 27,420 | 6,482 |
5.4 Regression Analysis Using Minitab
First, I used the Minitab DOE module to investigate the significance of factor interactions. The factorial design analysis (using α = 0.05) of the shrinkage volume results revealed that the main effects A and B were statistically significant (both P-values < 0.05), while the main effect C and all two-factor interactions were not significant (all P-values > 0.1). This justified building a main-effects-only linear regression model.
| Source | DF | Seq SS | Adj SS | Adj MS | F | P |
|---|---|---|---|---|---|---|
| Main Effects | 3 | 2,805,524 | 2,805,524 | 935,175 | 604.59 | 0.000 |
| A | 1 | 2,401,245 | 953,440 | 953,440 | 616.40 | 0.000 |
| B | 1 | 401,802 | 331,019 | 331,019 | 214.00 | 0.001 |
| C | 1 | 2,477 | 2,477 | 2,477 | 1.60 | 0.295 |
| 2-Factor Interactions | 3 | 6,396 | 6,396 | 2,132 | 1.38 | 0.399 |
| Residual Error | 3 | 4,640 | 4,640 | 1,547 | ||
| Total | 9 | 2,816,560 |
The regression equation obtained for the total shrinkage volume (Y) was:
$$Y = -56022 + 53.9A – 15.9B + 1.21C$$
with R² = 99.6% and P-value = 0.000, confirming excellent model fit. The F-test results are shown below.
| Source | DF | Seq SS | F |
|---|---|---|---|
| A | 1 | 2,401,245 | 1,305.73 |
| B | 1 | 401,802 | 218.49 |
| C | 1 | 2,477 | 1.35 |
The critical value is F₀.₀₅(1,6) = 5.99. Since F(3) = 1.347 < 5.99, factor C was not statistically significant and was removed from the model. The reduced model was:
$$Y = -54421 + 52.9A – 16.4B$$
with R² = 99.5%. The F-test for the reduced model was F(1) = 1244.17 and F(2) = 208.19, both exceeding the critical value F₀.₀₅(1,7) = 5.59. Therefore, both A and B significantly influence the total shrinkage volume, with pouring temperature being the dominant factor. Residual analysis confirmed that residuals were randomly distributed, validating the model choice.
To minimize Y, the equation dictates that A should be minimized and B maximized within the experimental domain. From the original full model, the optimal combination is:
$$Y_{\min} = -54421 + 52.9 \times 1550 – 16.4 \times 120 = 25{,}660 \text{ cc}$$
with the mold temperature at its minimum value of 10°C. This yields the optimal parameter set:
| Parameter | A (°C) | B (s) | C (°C) |
|---|---|---|---|
| Value | 1550 | 120 | 10 |
To confirm this optimum, I performed an additional simulation (Trial 11) with A = 1550°C, B = 120s, and C = 10°C. The result yielded a total shrinkage volume of 25.66 cc, which is indeed lower than all ten original trials, confirming the validity of the optimization.
For the Niyama shrinkage porosity response, the factorial analysis revealed that only factor B (pouring time) had a significant effect, while A and C showed negligible influence. Therefore, it was not meaningful to build a full regression model for the Niyama porosity response with these three factors. The Pareto chart and normal probability plot clearly indicated that only B was significant.
Chapter 6. Microstructure Study of ZG12MnMoV
6.1 Crystallization Process of ZG12MnMoV
ZG12MnMoV is a low-carbon (0.12% C) manganese-molybdenum-vanadium steel. According to the Fe-C phase diagram, the solidification path includes the peritectic reaction. The transformation sequence during cooling is:
L (liquid) → L + δ (ferrite) → L + δ + γ (austenite) → γ → γ + α (ferrite) → α + P (pearlite)
During primary crystallization, as the liquid cools below the liquidus line AB, delta ferrite nucleates. Upon reaching the peritectic temperature (HJB line), the peritectic reaction occurs, producing austenite from the liquid and delta ferrite. Austenite continues to form upon further cooling until the peritectic transformation is complete. During secondary crystallization, proeutectoid ferrite forms as the temperature falls below the GS line. Finally, at the eutectoid temperature (PS line), pearlite forms.
6.2 Dendritic Segregation
The peritectic reaction in this steel is often incomplete, leading to severe microsegregation after solidification. Dendritic segregation, a form of microsegregation, results from the partitioning of solute elements between dendrite arms and the interdendritic liquid during non-equilibrium solidification. This segregation significantly reduces the impact toughness and plasticity of the casting and increases the tendency for hot tearing. Consequently, a homogenization heat treatment is essential. The efficiency of such a treatment is strongly influenced by the dendrite arm spacing: the finer the dendritic structure (smaller secondary dendrite arm spacing, SDAS), the shorter the diffusion distance required for homogenization, and the more effective the heat treatment.
6.3 Procast Simulation and Analysis
I used the Micro module of Procast software to simulate the secondary dendrite arm spacing. The pouring temperature, which is the most influential macroscopic parameter, was selected as the study variable. Three levels were examined: 1550°C, 1565°C, and 1580°C, all within a reasonable processing range for this steel.
6.3.1 Mesh Division
The finite element model was created in the MeshCAST module of Procast. The casting model contained 445,088 elements and 89,219 nodes. A view of the mesh is shown in Figure 6.2 of the thesis.
6.3.2 Material Parameters
The main physical parameters of ZG12MnMoV used in the simulation are listed below.
| Property | Value |
|---|---|
| Density | 7.9 g/cm³ |
| Solidus Temperature | 1469°C |
| Liquidus Temperature | 1511°C |
The temperature-dependent properties including thermal conductivity, density, enthalpy, kinematic viscosity, and solid fraction were defined, and representative plots are given in the thesis. The boundary conditions were pouring temperature as listed above and a pouring velocity of 1.2 m/s.
6.3.3 Run Parameters
- Total number of calculation steps: 5000
- Stop criterion: temperature, set to 300°C
- Turbulence parameter: 1
- Micro module activated (MICRO=1), thermal module activated (therm=1)
6.3.4 Simulation Results and Analysis
Three simulations were run at pouring temperatures of 1550°C, 1565°C, and 1580°C. The simulation outputs for both the primary dendrite radius and the secondary dendrite arm spacing are shown in Figures 6.5 and 6.6 of the original thesis. The key finding was that as the pouring temperature increased, the primary dendrite radius and the secondary dendrite arm spacing both increased significantly throughout the casting, especially in the thickest sections—the bottom arc and the center region where the riser meets the base. The largest dendrite arm spacing correlated with coarser, less desirable microstructures. At 1550°C, both the primary dendrite radius and the SDAS were the smallest, and their distribution was the most uniform across the casting. This indicates that a lower pouring temperature refines the grain structure, reduces microsegregation, improves mechanical properties (notably impact toughness and plasticity), and reduces hot tearing tendency. The uniform SDAS distribution also suggests a more uniform overall mechanical response. Therefore, 1550°C was selected as the optimal pouring temperature, which is consistent with the macro-scale findings of the uniform design study. A lower pouring temperature also reduces the total heat input, which helps to minimize sand foundry defects by reducing mold-metal interaction time and severity.
Chapter 7. Conclusions and Future Perspectives
7.1 Main Conclusions
This research systematically addressed the sand foundry defects encountered during the production of a nuclear main pump bearing support casting. The key conclusions drawn from this work are summarized as follows:
(1) The original casting process, which positioned the base downward with small blind risers on the triangular ribs, was inadequate. The hot spot at the junction of the triangular rib and the vertical face had a significantly larger modulus than the risers provided, leading to insufficient feeding and extensive internal shrinkage defects. By inverting the casting orientation and installing a large insulating open riser (φ560 × 900 mm) directly above the base, combined with strategically placed chills to control the solidification sequence, I achieved complete feeding of the hot spots. Huazhu CAE simulation confirmed that the improved process eliminated all internal shrinkage cavity and porosity defects, and this was validated by MC2000 ultrasonic inspection in actual production.
(2) The severe sand burn-on defect at the massive hot spot was traced to the decomposition of zircon flour under prolonged high-temperature exposure. The SiO₂ released reacted with FeO to form low-melting fayalite slag, which consumed the coating and allowed direct metal penetration into the mold. My improvement—applying chromite sand facing locally at the hot spot and filling the remainder with furan resin sand, along with a thicker, more temperature-stable zircon coating applied in five coats—successfully eliminated the burn-on. This approach balanced technical effectiveness with economic efficiency.
(3) Uniform design provided a powerful and efficient method for optimizing the casting process parameters. Using only 10 simulation trials, I established that pouring temperature has the most significant influence on the total shrinkage volume, followed by pouring time, while the initial mold temperature exhibits minimal influence. The regression equation Y = -54421 + 52.9A – 16.4B accurately predicts the shrinkage volume within the studied domain. The optimal combination was found to be A = 1550°C, B = 120s, C = 10°C. A confirmatory simulation (Trial 11) produced a new minimum shrinkage volume of 25.66 cc, validating the predicted optimum. This combined uniform design and regression analysis methodology proved to be a robust and practical tool for process optimization in foundry engineering.
(4) The Procast Micro-module simulation demonstrated that the pouring temperature significantly affects the dendritic microstructure. Lower pouring temperatures within the investigated range (1550°C to 1580°C) yielded finer primary dendrites and smaller secondary dendrite arm spacing. At 1550°C, both the SDAS and primary dendrite radius were minimized, and their spatial distribution was most uniform, implying refined grains, reduced microsegregation, improved mechanical properties, and better homogenization response during subsequent heat treatment. The use of simulation tools for microstructural optimization complements the macroscopic defect analysis, providing a comprehensive approach to casting quality improvement.
7.2 Limitations and Future Research Directions
There are several limitations to this research that provide avenues for future exploration:
(1) The heat transfer coefficients at the casting-mold interfaces were treated as constant throughout the filling and solidification processes. In reality, these coefficients vary with temperature and pressure, resulting in a gap between simulated and real-world outcomes. Future work should incorporate temperature-dependent interface heat transfer models.
(2) The coating composition was optimized empirically in terms of application strategy and local sand selection. A more detailed study involving systematic variation of the coating component proportions, as well as other coating additives that might generate protective atmospheres, could yield further improvements.
(3) Microstructure simulation was limited to the secondary dendrite arm spacing. While instructive, other microstructural parameters such as primary dendrite arm spacing, grain size distribution, and the volume fraction of phases formed during the peritectic reaction would provide a more complete picture of the casting’s structure-property relationships.
(4) Future simulation studies should extend beyond SDAS to model the entire peritectic transformation explicitly, considering the effects of cooling rate, alloy composition, and convection on the final microstructure.
In summary, this research successfully resolved critical production problems around sand foundry defects, including shrinkage, porosity, and sand burn-on, through a combination of CAE simulation, optimized process design, coating technology improvement, statistical experimental design, and microstructural simulation. The resulting process improvements are producing castings of reliable quality suitable for demanding nuclear service. The methodologies developed here provide a valuable reference for both industrial foundry practice and academic research in the field. Further exploration is warranted to deepen the understanding of process-microstructure-property linkages and to further enrich the decision-making toolbox for advanced casting engineering.