This dissertation addresses critical manufacturing challenges encountered during the production of the CPR1000 nuclear main pump bearing support, a heavy-section casting weighing approximately 2.5 tonnes with wall thickness exceeding 200 mm. The component, made of ZG12MnMoV low-alloy steel, exhibited severe internal defects including shrinkage cavities, porosity, and sand adhesion during initial production trials. Given the demanding service environment of nuclear reactors—characterized by high pressure, elevated temperature, and intense radiation—the quality requirements for such castings are exceptionally stringent. This research systematically combines numerical simulation, experimental design methodology, and microstructural analysis to develop robust solutions for eliminating these defects. The primary technical approach involves using HuaZhu CAE software to simulate the filling and solidification processes, applying uniform design experiments with Minitab regression analysis to optimize process parameters, and employing Procast software to investigate microsegregation phenomena. The findings demonstrate that strategic modifications to the gating system, riser placement, chilling arrangement, and coating formulation effectively eliminate the observed defects, while optimum process parameters and pouring temperatures significantly improve casting quality and structural integrity.
1. Introduction and Research Significance
The nuclear main pump serves as the sole rotating equipment in the primary loop system of a pressurized water reactor (PWR), functioning continuously to circulate the reactor coolant. The pump bearing support is classified as a critical safety-related component, demanding exceptional reliability, leak-tightness, and durability throughout its design life. The casting operates under extreme environmental conditions including pressures exceeding 15 MPa, temperatures ranging from 300°C to 350°C, and exposure to intense neutron radiation. The base material specified for this application is 12MDV6 (domestic designation ZG12MnMoV), which belongs to the family of micro-alloyed cast steels containing vanadium, molybdenum, and manganese as primary alloying elements.
Micro-alloyed cast steels such as ZG12MnMoV exhibit superior combinations of strength, toughness, and weldability. The addition of vanadium and molybdenum promotes grain refinement through the precipitation of stable carbides and nitrides, while manganese contributes to solid solution strengthening. However, the solidification characteristics of these steels present considerable challenges. The composition places this alloy within the peritectic region of the Fe-C phase diagram, leading to complex phase transformations during solidification. The peritectic reaction between delta ferrite and liquid produces austenite, and this transformation is inherently incomplete, resulting in significant microsegregation within the dendritic structure. This microsegregation compromises mechanical properties, reduces corrosion resistance, and increases susceptibility to hot tearing.
The foundry industry has witnessed remarkable advancements in computer-aided engineering (CAE) technologies for casting process simulation. These tools enable foundry engineers to visualize and analyze filling patterns, solidification sequences, temperature distributions, and defect formation mechanisms before physical trial production. The application of such simulation technologies significantly reduces development costs, shortens lead times, and improves first-time-right ratios. Unlike traditional trial-and-error approaches, simulation-based optimization allows for systematic exploration of process parameter spaces and provides quantitative insights into complex physical phenomena governing casting quality.
The challenges associated with heavy-section castings are particularly demanding due to the extended solidification times, large thermal gradients, and increased susceptibility to feeding deficiencies. For the main pump bearing support, the combination of heavy section thickness and complex geometry creates multiple hot spots where shrinkage defects preferentially form. The presence of sand foundry defect such as metal penetration and burn-on further complicates the manufacturing process, particularly in regions where prolonged high temperatures overcome the protective capabilities of conventional mold coatings.
2. Casting Characteristics and Initial Process Design
2.1 Component Geometry and Material Specification
The main pump bearing support exhibits a complex geometry with an overall envelope of 1000 mm × 1000 mm × 1103 mm. The wall thickness remains remarkably uniform throughout the component, consistently exceeding 200 mm. The casting mass is approximately 2.5 tonnes, classifying it as a heavy-section casting requiring careful thermal management during solidification. The design incorporates multiple structural features including a base plate, vertical walls, triangular reinforcing ribs, circular bosses, and an arched top section. The intersection between the triangular ribs and the vertical walls creates a significant local thickening that becomes a preferential site for shrinkage defect formation.
Table 2.1 presents the chemical composition requirements for ZG12MnMoV according to the domestic standard NB_T 20008.2-2010.
| Element | C | Si | Mn | P | S | Mo | V |
|---|---|---|---|---|---|---|---|
| Content (wt%) | ≤0.15 | ≤0.60 | 1.20–1.70 | ≤0.025 | ≤0.020 | 0.20–0.40 | 0.05–0.10 |
The mechanical property requirements ensure adequate strength and ductility for structural applications in nuclear environments, as summarized in Table 2.2.
| Property | Yield Strength σs (MPa) | Tensile Strength σb (MPa) | Elongation δ (%) |
|---|---|---|---|
| ZG12MnMoV | ≥400 | ≥500 | ≥18 |
The manganese content, while beneficial for strength, promotes grain coarsening tendencies. Combined with the peritectic solidification behavior, this necessitates careful process design to achieve the required ultrasonic testing standards (MC 2000) for internal soundness.
2.2 Initial Process Parameters
The original manufacturing process employed a resin-bonded sand mold system with furan resin as the binder. The molding method was manual sand molding utilizing a four-part molding box system. The initial process configuration incorporated a stepped gating system constructed from refractory fireclay tubes. The key dimensional parameters of the original gating system are: sprue diameter 80 mm, runner diameter 80 mm, ingate diameter 50 mm, and ladle nozzle diameter 45 mm. The pouring cup was designed with an 80 mm diameter opening.
The initial riser system comprised one insulated blind riser of waist-circular cross-section positioned at the arched top of the casting, designated as riser #1 with dimensions a=200 mm, b=300 mm, and h=300 mm, with a sleeve thickness of 30 mm. Additionally, two circular insulated blind risers were placed on the triangular ribs (riser #2), each with a diameter of 200 mm and sleeve thickness of 15 mm. To enhance directional solidification, twelve external chills were incorporated: two bar-type chills measuring 700×50×50 mm positioned symmetrically on the thick wall adjacent to the triangular ribs, and eleven rectangular chills measuring 200×160×80 mm distributed on the triangular rib sides and beneath the circular base.
The pouring temperature was set at 1550°C with a filling time of 90 seconds. The casting was subjected to a quench and temper heat treatment after solidification and cooling. A water-based molding coating with zircon flour as the refractory filler was applied by brush coating in a single layer.
3. Initial Process Simulation and Defect Analysis
3.1 Simulation Methodology
The three-dimensional solid models of the casting, gating system, risers, chills, and mold were constructed using Unigraphics (UG) software and exported in STL file format. These files were subsequently imported into the HuaZhu CAE simulation environment for mesh generation and analysis. The priority order for material assignment was established as casting, core, chill, riser sleeve, and mold, in descending order of priority.
For pure solidification calculations, a uniform grid size of 7.7 mm was selected, resulting in approximately 8.17 million grid elements. This fine resolution ensures accurate thermal gradient predictions essential for defect analysis. For coupled filling and solidification calculations, the grid count was reduced to approximately 600,000 elements to maintain computational efficiency while preserving calculation accuracy.
The simulation parameters for ZG12MnMoV included: liquidus temperature 1511°C, solidus temperature 1469°C, and pouring temperature 1550°C. The gravity-assisted feeding option was activated to enable quantitative prediction of shrinkage porosity. The calculation termination criterion was set at 96% solidification of the casting, with automatic data saving every 100 time steps.
3.2 Filling Process Analysis
Temperature field evolution during mold filling provides critical insights into heat transfer patterns and identifies potential areas of premature cooling or excessive heating. The simulation results at various filling stages revealed distinct thermal behaviors:
During the initial 15% to 50% of filling, the lowest temperature regions (1515°C) appeared at locations adjacent to the chilling elements near the base, demonstrating effective heat extraction by these cooling devices. Conversely, the central base area maintained higher temperatures due to the cumulative thermal mass of accumulating molten metal. This gradient intensified as filling progressed, establishing the foundation for directional solidification. At 87% to 96% filling, the minimum temperature (1512°C) shifted to regions influenced by the chills on the vertical walls, indicating localized accelerated cooling that potentially interferes with riser feeding efficiency.
Flow field analysis demonstrated appropriate flow characteristics with relatively calm progression through the mold cavity. However, several concerns emerged: higher flow velocities at the ingate-casting junction and at the transition between the triangular ribs and risers create potential for mold erosion and reoxidation. These turbulence zones increase the likelihood of gas entrainment and non-metallic inclusion entrapment, both contributing to quality degradation.
3.3 Solidification Analysis and Defect Identification
The pure solidification calculation provided comprehensive insight into the solidification sequence and defect formation mechanisms. Figure-based analysis of the temperature distribution at various time steps (476.95 s, 1927.53 s, 4905.79 s, and 11858.87 s) revealed the progressive solidification pattern. The ideal directional solidification directed toward the top risers was not fully achieved due to insufficient riser capacity at critical locations.
The simulation results identified substantial shrinkage cavities within the triangular rib regions, as revealed in the final solidification stage. This deficiency indicates that the circular blind risers on the triangular ribs possessed insufficient modulus to maintain feed paths until solidification completion. The volumetric shrinkage of the steel during solidification exceeded the available liquid metal supply from the inadequate risers, resulting in internal cavity formation. The quantification of defects confirmed that the original process could not adequately feed the hot spot at the junction between triangular ribs and vertical walls.
The simulation also revealed that the bar-type chills exceeded the recommended length-to-thickness ratio, increasing their susceptibility to deformation with repeated use. Such deformation compromises both dimensional accuracy and chilling efficiency, further contributing to localized solidification irregularities.
4. Process Improvement and Validation
4.1 Modified Process Design
Based on the comprehensive defect analysis, the casting process was redesigned with several strategic modifications. The fundamental change involved inverting the casting orientation—positioning the large base plate at the top of the mold to facilitate placement of larger, more effective risers directly on the major hot spots. This inverted orientation creates natural feeding paths aligned with gravity, enhancing the efficiency of liquid metal delivery to solidification shrinkage zones.
Table 4.1 summarizes the key modifications between the original and improved processes.
| Parameter | Original Process | Improved Process |
|---|---|---|
| Orientation | Base plate down | Base plate up |
| Riser type | Blind insulated risers | Open atmospheric risers |
| Riser dimensions | 200×300 mm, φ200 mm | φ560×900 mm |
| Riser sleeve thickness | 15–30 mm | 112 mm |
| Chill configuration | Bar 700×50×50, Rectangular 200×160×80 | Cylindrical φ30×70 (24 pcs), Bar 180×120×90 (10 pcs) |
| Pouring temperature | 1550°C | 1550–1570°C |
| Filling time | 90 s | 80–120 s |
The modified gating system maintained a stepped configuration but with repositioned ingates to achieve more uniform flow distribution. The riser system was completely redesigned with a single large open atmospheric riser positioned on the base plate, providing substantial liquid metal reservoir with enhanced feeding pressure. The atmospheric riser design maintains an open hole through the mold, ensuring the feeding channel remains clear throughout solidification and enabling visual inspection of filling level.
The chilling strategy was refined to more effectively manage local solidification rates. A total of 24 cylindrical chills (φ30×70 mm) were positioned along the fillet radii at the triangular rib-wall intersections, directly targeting the identified hot spot regions. Additionally, 10 bar-type chills (180×120×90 mm) were distributed on the four circular bosses and arched top section. This chill arrangement accelerates solidification at strategic locations, increasing the effective feeding distance of the top riser and promoting more uniform progressive solidification.
4.2 Modified Process Simulation Results
4.2.1 Filling Behavior
Temperature field analysis for the modified process demonstrated more controlled thermal gradients throughout the filling sequence. At 10% filling, the minimum temperature of 1527°C appeared at the arched bottom region under the influence of the newly positioned chills. The gradual temperature decrease during filling (reaching 1493°C at 98% filling on the boss areas) indicates effective chill performance in creating localized cooling zones that guide solidification direction. The overall temperature drop remained modest, confirming adequate superheat retention and minimizing the risk of cold shut defects.
Flow visualization revealed smooth, progressive mold filling without excessive turbulence. The metal front advanced uniformly through the cavity, with the arch section filling first, followed by the vertical walls and base section. No evidence of short pouring or mistun defects was observed. The refined flow patterns reduced mold erosion potential and facilitated the flotation of inclusions into the top riser.
4.2.2 Solidification Analysis
The solidification simulation for the modified process demonstrated significantly improved feeding behavior. The temperature distribution at comparable time steps showed a more coherent solidification front advancing from the chilled regions toward the top riser. Crucially, the hot spot regions at the triangular rib-wall intersections remained connected to liquid metal through established feeding channels until solidification completion.
The final solidification result exhibited remarkable improvement—the shrinkage cavities present in the original process were completely eliminated. The large open riser provided adequate liquid metal supply with sufficient feeding pressure to compensate for solidification shrinkage throughout the casting. The combination of strategic chilling and enhanced riser feeding achieved the desired progressive solidification pattern essential for producing sound castings.
The modified process was implemented for production validation. The produced casting displayed a clean external surface with no visible shrinkage defects. Ultrasonic examination per MC 2000 standards confirmed the casting met all internal quality requirements with dense, uniform structure. The dimensional inspection passed all tolerance requirements, confirming the effectiveness of the process modifications.
4.3 Sand Foundry Defect Prevention Through Coating Optimization
Despite the successful elimination of shrinkage defects, the production trial revealed significant sand adhesion at the triangular rib-wall intersection—precisely where substantial thermal mass created prolonged high-temperature conditions. This sand foundry defect manifested as thick layers of fused sand strongly bonded to the casting surface, presenting severe cleaning difficulties and compromising surface quality.
4.3.1 Mechanism of Sand Foundry Defect at Hot Spots
The sand adhesion mechanism at the critical hot spot differs fundamentally from that at other casting locations. The zircon flour-based coating, while generally effective, has a decomposition temperature around 1540°C. At the triangular rib-wall intersection, the extremely heavy section causes the adjacent mold surface to remain above this critical temperature for extended periods. Under these prolonged thermal conditions, zircon flour decomposes into silica (SiO₂) and zirconia (ZrO₂):
$$ \text{ZrSiO}_4 \xrightarrow{\Delta T} \text{ZrO}_2 + \text{SiO}_2 $$
The molten steel simultaneously oxidizes at the metal-mold interface, forming iron oxide (FeO):
$$ 2\text{Fe} + \text{O}_2 \rightarrow 2\text{FeO} $$
These oxides subsequently react to form low-melting-point fayalite:
$$ \text{FeO} + \text{SiO}_2 \rightarrow \text{FeSiO}_3 \quad (\text{melting point} \approx 1200°C) $$
This compound flows into the porous mold surface, completely destroying the protective coating layer. Once the coating is breached, molten steel penetrates directly into the sand mold pore structure, forming a deeply embedded mechanical penetration layer that is extremely difficult to remove. At other casting locations, faster cooling maintains the interface temperature below the decomposition threshold, allowing zircon to undergo solid-state sintering that forms a protective ceramic barrier.
4.3.2 Improved Coating and Sand System
To address this severe sand foundry defect, a multi-pronged coating strategy was developed. The approach combines localized use of chromite sand as facing sand at the critical hot spot regions with optimized zircon-based coating applied uniformly across the mold cavity. Chromite sand (FeO·Cr₂O₃) offers superior thermal properties with a melting point exceeding 1900°C. During contact with molten steel, chromite sand undergoes a unique solid-state transformation where the sand grains increase in volume and deform plastically, transitioning from point contact to face contact:
| Property | Zircon Sand | Chromite Sand |
|---|---|---|
| Chemical composition | ZrSiO₄ | FeO·Cr₂O₃ |
| Melting point (°C) | ~1850 | >1900 |
| Thermal conductivity | Moderate | High |
| Decomposition temperature (°C) | ~1540 | Stable to >1700 |
| Volume stability at high T | Decomposes | Expands, seals pores |
This grain expansion behavior effectively seals interstices between sand particles, creating an impermeable barrier against metal penetration. The chromite sand facing layer was applied with thickness of 40–50 mm at the critical triangular rib-wall junction locations, with furan resin-bonded sand as backing material.
The coating formulation was enhanced by replacing the original sugar syrup binder with a combination of water-soluble phenolic resin (for room temperature strength) and silica sol (for high-temperature bonding). The refractory filler remained as zircon flour due to its excellent overall performance characteristics. The coating thickness was increased substantially by applying 5 brush coats instead of 1, achieving a final dry film thickness of 1.2–1.5 mm.
Production validation of the improved coating system confirmed complete elimination of sand adhesion defects at the hot spot regions. The casting surface quality improved dramatically, significantly reducing post-casting cleaning time and labor costs. The localized chromite sand application minimized cost increases while achieving the required quality improvements. Figure 6 displays the successfully produced casting, demonstrating effective implementation of the optimizations.
5. Uniform Design Optimization of Process Parameters
5.1 Uniform Design Methodology
Uniform design, developed by Chinese mathematicians, provides an efficient experimental methodology for optimizing multi-factor systems. Unlike orthogonal designs that require meeting the “uniform dispersion and comparable alignment” criteria, uniform design emphasizes only the uniform distribution of experimental points within the test domain. This approach dramatically reduces the required number of experiments while extracting maximum information from each trial. The theoretical foundation of uniform design rests on number-theoretic methods in numerical integration and quasi-Monte Carlo techniques.
For a system with m factors each having n levels, complete factorial experimentation requires n^m experiments. Orthogonal design reduces this to 2n experiments, while uniform design needs only n experiments. This efficiency makes uniform design particularly valuable for optimization problems with multiple factors and many levels, such as casting process parameter optimization.
5.2 Experimental Design
Three critical process parameters were selected for optimization: pouring temperature (A), pouring time (B), and initial mold temperature (C). Each factor was assigned five experimental levels within industrially relevant ranges: pouring temperature from 1550°C to 1570°C, pouring time from 80 to 120 seconds, and mold initial temperature from 10°C to 50°C.
Based on the uniform design table U10*(10^8), which features superior uniformity compared to non-starred design tables, columns 1, 5, and 6 were determined most appropriate for accommodating the three factors while achieving optimal experimental point distribution. The resulting experimental scheme is presented in Table 5.6.
| Experiment | A – Pouring Temp (°C) | B – Pouring Time (s) | C – Mold Temp (°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 |
Each experimental condition was simulated using HuaZhu CAE software with consistent mesh parameters and boundary conditions. Two response variables were extracted: total shrinkage cavity volume from the vol% display and Niyama shrinkage porosity values. These responses provide quantitative measures of casting soundness.
5.3 Regression Analysis for Shrinkage Cavity Volume
The simulation results for total shrinkage cavity volume across the ten experiments are compiled in Table 5.8.
| Experiment | A (°C) | B (s) | C (°C) | Shrinkage Volume (cc) |
|---|---|---|---|---|
| 1 | 1550 | 120 | 20 | 25,680 |
| 2 | 1555 | 120 | 30 | 25,940 |
| 3 | 1560 | 110 | 50 | 26,370 |
| 4 | 1565 | 110 | 10 | 26,640 |
| 5 | 1570 | 100 | 20 | 26,980 |
| 6 | 1550 | 100 | 40 | 25,960 |
| 7 | 1555 | 90 | 50 | 26,430 |
| 8 | 1560 | 90 | 10 | 26,620 |
| 9 | 1565 | 80 | 30 | 27,140 |
| 10 | 1570 | 80 | 40 | 27,420 |
Preliminary factorial analysis using Minitab DOE module revealed that the main effects of factors A and B were statistically significant (P-values less than 0.05), while factor C showed no significant effect. The two-factor interactions were found statistically negligible, enabling simplification of the regression model to a first-order linear form.
Multiple linear regression yielded the following equation:
$$ Y = -56022 + 53.9A – 15.9B + 1.21C $$
where Y represents the total shrinkage cavity volume. The coefficient of determination R² = 99.6% indicates excellent model fit. However, since factor C proved statistically insignificant (P = 0.290), the model was refined by removing this term:
$$ Y = -54421 + 52.9A – 16.4B $$
Table 5.13 summarizes the analysis of variance (ANOVA) for this refined model.
| Source | DF | Seq SS | Adj SS | Adj MS | F | P |
|---|---|---|---|---|---|---|
| Regression | 2 | 2,803,047 | 2,803,047 | 1,401,523 | 726.00 | 0.000 |
| A | 1 | 2,401,245 | — | — | 1,244.17 | <0.001 |
| B | 1 | 401,802 | — | — | 208.19 | <0.001 |
| Residual Error | 7 | 13,513 | — | 1,930 | — | — |
| Total | 9 | 2,816,560 | — | — | — | — |
F-test comparison against tabulated critical values F(0.05, 1, 7) = 5.59 confirms both factors significantly influence the response. The influence ranking is A (pouring temperature) having the greatest impact, followed by B (pouring time). Residual analysis confirmed the model adequacy with random scatter patterns in the residuals versus order plot.
According to the regression equation, minimizing shrinkage cavity volume requires taking the lowest pouring temperature (1550°C) and the longest pouring time (120 s) within the experimental ranges. To verify this prediction, an additional confirmation experiment (experiment 11) was conducted with pouring temperature 1550°C, pouring time 120 s, and mold temperature 10°C—the optimal combination including the recommended lowest mold temperature from the full model. The simulated shrinkage cavity volume of 25.66 cc was indeed lower than all previous experiments, confirming the optimization accuracy.
5.4 Niyama Shrinkage Porosity Analysis
Niyama criterion provides an alternative indicator for assessing micro-porosity risk based on the local thermal gradient (G) and cooling rate (R). The Niyama parameter N_y is defined as:
$$ N_y = \frac{G}{\sqrt{R}} $$
Lower Niyama values indicate higher susceptibility to micro-porosity formation. The simulation results for Niyama shrinkage porosity across the ten uniform design experiments showed different factor significance patterns compared to the total shrinkage volume. Pareto analysis revealed that only pouring time (factor B) exhibited statistically significant influence on Niyama porosity values. Both pouring temperature and mold temperature showed negligible effects within the tested ranges.
The data distribution from the simulation results indicates that Niyama porosity values ranged from 4,432 cc (experiment 3) to 7,157 cc (experiment 6), showing considerable scatter without clear linear relationships with pouring temperature or mold temperature. This finding suggests that micro-porosity formation in this heavy-section casting is more strongly governed by solidification rate and thermal gradient evolution than by initial process parameters.
6. Microstructural Analysis of ZG12MnMoV
6.1 Solidification Path and Microsegregation Origin
The ZG12MnMoV alloy, as a low-carbon micro-alloyed steel, undergoes a characteristic solidification path. Referencing the Fe-C phase diagram, the cooling sequence from liquid to ambient temperature involves: liquid → δ-ferrite + liquid → δ-ferrite + austenite → austenite → austenite + ferrite → pearlite + ferrite. The transformation passes through the peritectic region where significant line contraction occurs during the peritectic reaction between δ-ferrite particles and remaining liquid to form austenite:
$$ L + \delta\text{-Fe} \xrightarrow{1495°C} \gamma\text{-Fe} $$
The peritectic reaction is diffusion-controlled and frequently incomplete in practice, creating compositional heterogeneities at the micro-scale. During subsequent cooling through the austenite plus ferrite region, diffusion becomes increasingly sluggish, effectively “freezing in” these compositional gradients. This phenomenon creates the characteristic dendritic segregation pattern where solute elements such as manganese and molybdenum concentrate in inter-dendritic regions while dendrite cores become depleted. Such microsegregation compromises the homogenization efficiency of subsequent heat treatments.
6.2 Simulation Methodology Using Procast Micro Module
To quantify microsegregation, the secondary dendrite arm spacing (SDAS) was selected as the key microstructural parameter. Smaller SDAS values correlate with finer structures, more uniformly distributed solute, and superior mechanical properties. Procast software’s Micro module was employed to simulate dendrite evolution during solidification.
The finite element mesh was generated in MeshCAST with 445,088 elements and 89,219 nodes. Material properties including thermal conductivity, density, enthalpy, and kinematic viscosity were defined as temperature-dependent functions to accurately capture solidification phenomena. The thermal-physical property distributions are presented in Figure 6.3.
The boundary conditions incorporated a pouring temperature of 1550°C and filling velocity of 1.2 m/s. The Micro calculation module was activated together with the thermal simulation. Three pouring temperatures (1550°C, 1565°C, and 1580°C) were investigated to assess the influence of superheat on dendrite morphology.
6.3 Results and Discussion
The simulation results, presented in Figures 6.5 and 6.6, revealed a direct relationship between pouring temperature and dendrite dimensions. At 1550°C, the primary dendrite arm spacing and secondary dendrite arm spacing showed minimum values across all casting sections. The hierarchical equation governing dendritic growth rate dictates that smaller constitutional supercooling at lower pouring temperatures inhibits dendrite coarsening:
Table 6.3 compares representative SDAS values at different pouring temperatures.
| Pouring Temperature (°C) | Secondary Dendrite Arm Spacing (μm) | Relative Fineness |
|---|---|---|
| 1550 | Reference (minimum) | Best |
| 1565 | ~25% increase | Moderate |
| 1580 | ~45% increase | Coarsest |
With increasing pouring temperature, the secondary dendrite arm spacing substantially increased, particularly at the bottom arch region and the riser-casting interface at the top. The coarser dendritic structure exacerbates microsegregation by increasing solute diffusion distances. This parameter directly affects the effectiveness of homogenization heat treatment:
$$ t \propto \frac{\lambda_2^2}{D} $$
where λ₂ is secondary dendrite arm spacing, D is the diffusion coefficient, and t is the required homogenization time. The practical consequence of this relationship is that castings produced at higher pouring temperatures require substantially longer heat treatment cycles to achieve equivalent microstructure uniformity.
The simulations at 1550°C demonstrated not only the finest dendrites but also relatively uniform SDAS distribution throughout the casting, indicating more consistent microstructural properties across different sections. The improved homogeneity facilitates processing and ensures dependable service performance. Based on these findings, 1550°C was confirmed as the optimal pouring temperature, balancing the competing requirements of fluidity during mold filling and microstructural refinement during solidification.
7. Conclusions and Future Perspectives
7.1 Summary of Achievements
This research successfully addressed multiple manufacturing challenges associated with producing the CPR1000 nuclear main pump bearing support in ZG12MnMoV micro-alloyed cast steel. The comprehensive investigation combining simulation, experimental design, and production validation delivered the following key conclusions:
Firstly, the primary cause of shrinkage defects in the original process was attributed to insufficient riser modulus at the triangular rib hot spots. Under-sized blind risers could not maintain feeding paths until solidification completion. The modified process, incorporating an inverted casting orientation and a large open atmospheric riser (φ560×900 mm) with 112 mm insulation sleeve, achieved complete soundness across all casting sections, as confirmed by MC 2000 ultrasonic testing.
Secondly, the interplay between localized heavy sections and coating performance determines the susceptibility to sand foundry defect formation. The decomposition of zircon flour above 1540°C at long-duration hot spots leads to coating failure and subsequent metal penetration. The strategic incorporation of chromite sand facing (40–50 mm thickness) in these critical regions, combined with increased coating thickness (1.2–1.5 mm) and improved binder formulation, successfully eliminated sand foundry defect at the triangular rib-wall intersections. This solution substantially reduced cleaning costs while maintaining economic viability through localized sand application.
Thirdly, uniform design experimentation combined with Minitab regression analysis provided efficient process optimization with only eleven simulations. The analysis established that pouring temperature exerts the dominant influence on shrinkage cavity volume (R² = 99.5%), followed by pouring time, while mold initial temperature showed negligible effect within the tested range. The optimal parameter combination of 1550°C pouring temperature, 120 s pouring time, and 10°C mold temperature produced minimal shrinkage cavity volume, verified through additional simulation. This methodology demonstrates significant potential for accelerating process development in heavy-section steel castings.
Finally, Procast Micro module simulations revealed that pouring temperature significantly affects dendritic refinement through the secondary dendrite arm spacing. Lower pouring temperature (1550°C) produced the finest and most uniform dendrite structure, promoting easier homogenization of microsegregation during subsequent heat treatment. This resulted in improved mechanical properties, enhanced toughness, and reduced susceptibility to hot tearing.
7.2 Limitations and Future Research Directions
Several limitations of the current study suggest directions for future investigations. The simulation software assumed constant interfacial heat transfer coefficients throughout the casting process. In reality, these coefficients vary dynamically with temperature distributions, interfacial pressures, and the evolution of air gaps due to casting contraction and mold expansion. Implementation of temperature-dependent and pressure-dependent heat transfer models would enhance prediction accuracy.
The coating optimization primarily focused on empirical improvements and verification through production trials. Future research could systematically investigate the influence of individual coating constituents—refractory filler type and size distribution, binder ratios, and additives—on the high-temperature behavior and penetration resistance at simulated interfacial conditions. This would enable more rational coating design tailored to specific casting geometries and thermal conditions.
The microstructural simulation addressed secondary dendrite arm spacing as a proxy for microsegregation. However, more sophisticated multi-component solidification models incorporating diffusion kinetics and peritectic reaction dynamics could provide quantitative predictions of actual compositional profiles. Coupling such models with precipitation and homogenization simulation would enable end-to-end prediction of final mechanical properties from process parameters, facilitating more comprehensive optimization.
Future work could also extend the uniform design approach to include additional parameters such as gating system dimensions, chill placement configurations, and mold materials. Kriging metamodels or neural network surrogates trained on simulation data could enable rapid multi-objective optimization that considers competing quality metrics simultaneously, including soundness, productivity, and cost.
The integration of machine learning with casting simulation represents a promising frontier. Comprehensive databases of simulation results could enable predictive models that identify optimal process windows without requiring expensive iterative simulations. This approach would be particularly valuable for nuclear components where stringent quality requirements must be balanced with cost-effective production.
The successful elimination of the sand foundry defect at hot spot regions and internal shrinkage cavities in the main pump bearing support demonstrates the power of combining simulation-driven analysis with systematic experimental design. The methodology established in this research offers a robust framework for optimizing complex heavy-section casting processes, contributing to improved reliability and efficiency in nuclear component manufacturing.