1. Introduction to Casting Simulation and Sand Foundry Defect Control
The foundry industry plays a vital role in the manufacturing sector, contributing significantly to national economic development. Historically, casting technology has relied heavily on empirical experience passed down through generations of foundry engineers. However, the casting process represents an extremely complex physicochemical phenomenon involving multiple disciplines such as physics, fluid mechanics, heat transfer, metallurgy, and mechanics. The scientific analysis of casting solidification processes demands sophisticated experimental observation and mathematical analysis tools, which have become increasingly available with the advancement of computer technology.
Computer numerical simulation technology has emerged as a powerful approach to address the challenges of sand foundry defect prediction and quality control. By simulating the casting solidification process, engineers can visualize temperature distributions, predict shrinkage porosity and other defects before actual production, thereby optimizing casting processes and reducing trial-and-error costs. The application of computer technology in foundry engineering has become essential for producing high-quality castings and represents the primary direction of modernization in the casting industry.
The development of casting process simulation systems has progressed through several stages over the past four decades. Initially, researchers focused on macro-scale heat transfer analysis during solidification. Subsequent developments incorporated solidification kinetics and micro-scale simulation of liquid-solid coupling phenomena. Most recently, research efforts have concentrated on atomic-scale modeling. Modern casting simulation systems typically encompass four major components: mold filling simulation (flow field), solidification simulation (temperature field), thermal stress and residual stress simulation (stress field), and microstructure simulation (microstructure field).
Among these, temperature field simulation during solidification is the most mature and widely applied technology. This approach enables the determination of critical parameters such as temperature gradients, solid fractions, and solidification times, which serve as the basis for predicting shrinkage cavities and porosity. The accurate prediction of these sand foundry defects is essential for optimizing casting designs and improving product quality.
To fully leverage the capabilities of numerical simulation, there is a pressing need to develop specialized systems that integrate casting domain knowledge with general-purpose finite element analysis software. This thesis presents the design and development of a comprehensive system for predicting shrinkage cavity and porosity defects in castings, along with an intelligent defect analysis subsystem based on expert system technology. The system is built on the ANSYS platform, utilizing its parametric design language and user interface design language as internal development tools, combined with external development tools for database management and user interface enhancement.
2. Theoretical Foundations of Solidification Temperature Field Simulation
2.1 Heat Transfer Mechanisms and Governing Equations
The solidification process of castings involves complex heat transfer phenomena, primarily categorized into three modes: conduction, convection, and radiation. For a closed system, the energy conservation equation follows the first law of thermodynamics:
$$Q – W = \Delta U + \Delta KE + \Delta PE$$
For most engineering heat transfer problems, kinetic and potential energy changes are negligible, and no work is performed. Thus, the equation simplifies to:
$$Q = \Delta U$$
When heat flow rates are constant with respect to time, steady-state thermal analysis is appropriate. However, casting solidification is inherently a transient process where temperatures, heat flow rates, and boundary conditions change significantly with time. The transient heat balance matrix equation is given by:
$$[C]\{\dot{T}\} + [K]\{T\} = \{Q\}$$
where [C] is the specific heat matrix, [K] is the conduction matrix, {T} is the nodal temperature vector, and {Q} is the nodal heat flow rate vector. When material properties vary with temperature, boundary conditions depend on temperature, or radiation heat transfer is considered, the analysis becomes nonlinear:
$$[C(T)]\{\dot{T}\} + [K(T)]\{T\} = \{Q(T)\}$$
For the casting solidification process, the three-dimensional transient heat conduction equation, incorporating latent heat release, is expressed as:
$$\rho c \frac{\partial T}{\partial t} = \frac{\partial}{\partial x}\left(k_x \frac{\partial T}{\partial x}\right) + \frac{\partial}{\partial y}\left(k_y \frac{\partial T}{\partial y}\right) + \frac{\partial}{\partial z}\left(k_z \frac{\partial T}{\partial z}\right) + \dot{q}$$
where ρ, c, kx, ky, kz represent density, specific heat, and thermal conductivities in x, y, and z directions, respectively, and q̇ represents the latent heat release rate per unit volume.
2.2 Shrinkage Characteristics of Casting Alloys
The volume reduction experienced by molten metal from liquid state to solid state at room temperature is termed shrinkage. This shrinkage occurs in three distinct stages: liquid shrinkage, solidification shrinkage, and solid-state shrinkage. The liquid shrinkage coefficient can be expressed as:
$$\varepsilon_{VL} = \alpha_{VL}(T_{p} – T_{L})$$
where αVL is the liquid shrinkage coefficient, Tp is the pouring temperature, and TL is the liquidus temperature. Liquidation and solidification shrinkage constitute the fundamental causes of shrinkage cavity formation in castings. The volume of shrinkage cavity can be calculated as:
$$V_s = V_0 (\varepsilon_{L} + \varepsilon_{S}) – V_p (\varepsilon_{S} + \varepsilon_{P})$$
where V0 is the initial liquid metal volume, Vp represents the volume reduction due to mold expansion, and εL, εS, εP are liquid, solidification, and mold expansion coefficients.
2.3 Solid Fraction and Phase Transformation Analysis
The solid fraction is a critical parameter in solidification analysis. For practical multi-component alloys, the relationship between solid fraction and temperature can be determined using thermal analysis methods. In this system, we propose two distribution models for the solid fraction during solidification:
Linear distribution model: The solid fraction increases linearly as temperature decreases:
$$f_s = \frac{T_L – T}{T_L – T_S}$$
Parabolic distribution model: The solid fraction follows a parabolic relationship with temperature:
$$f_s = \left(\frac{T_L – T}{T_L – T_S}\right)^2$$
Phase transformation analysis represents a special case of transient thermal analysis where the latent heat released during solidification must be accounted for. Several methods exist for handling latent heat, including the temperature recovery method, equivalent specific heat method, and enthalpy method. The equivalent specific heat method is expressed as:
$$c_{eq} = c – L \frac{\partial f_s}{\partial T}$$
where L is the latent heat of solidification. This system employs the enthalpy method, which defines the enthalpy as the integral of heat capacity with respect to temperature, providing a natural and robust treatment of latent heat effects.
2.4 Critical Solid Fraction in Sand Foundry Defect Prediction
The critical solid fraction concept is fundamental to predicting sand foundry defects such as shrinkage cavities and porosity. When the solid fraction exceeds a critical value, liquid metal flow becomes extremely difficult, and volumetric contraction can no longer be compensated by liquid metal feeding. This critical solid fraction, denoted as fsc, depends on alloy composition and solidification morphology. Table 1 presents critical solid fractions for various alloys used in foundry practice.
| Alloy | Critical Solid Fraction | Remarks |
|---|---|---|
| Fe-C alloy (2.5%C) | 0.55±0.05 | Considering concentration after solidification |
| Fe-C alloy (4.3%C) | 0.9±0.05 | Measured concentration values and calculated by lever rule |
| Fe-C alloy (5.0%C) | 0.85±0.03 | Calculated using Flemings formula |
| Steel (0.18%C) | 0.75-0.8 | Metal mold, calculated using Flemings formula |
| Steel (0.18%C) | 0.75-1.0 | Metal mold, calculated using Flemings formula |
| Nodular cast iron | 0.65-0.7 | No clear basis given |
| Al-Cu alloy (4.5%Cu) | 0.9 | Free solidification, based on measurements |
| Al-Si alloy (7%Si) | 0.85 | Based on measurements of solid fraction and liquid flow |
| Cast steel (0.3%C) | 0.7-0.8 | Production experience |
| Stainless steel | 0.7-0.8 | Production experience |
3. System Architecture and Functionality
3.1 Overall System Design
The casting defect prediction and analysis system is designed as a comprehensive solution that integrates several functional subsystems. The hierarchical structure of the system is illustrated in Table 2, which outlines the primary subsystems and their core functions.
| Subsystem | Core Functions |
|---|---|
| Casting Process CAD | Casting design, riser design, mold design, gating system design |
| Temperature Field Simulation | Material property database, mesh generation, solution computation, result visualization |
| Defect Prediction | Temperature gradient method, Niyama criterion method, defect distribution visualization |
| Defect Analysis | Defect classification, cause analysis, prevention measures, knowledge base management |
3.2 Casting Process CAD Subsystem
The casting design module determines part orientation, parting surface selection, casting shrinkage allowance, machining allowance, and draft angles. The riser design module establishes riser type, quantity, position, and dimensions based on modulus calculations and feeding distance analysis. Table 3 summarizes the material databases utilized in this subsystem.
| Database Name | Contents |
|---|---|
| Casting alloy linear shrinkage database | Linear shrinkage coefficients for various alloys |
| Machining allowance database | Recommended machining allowances based on alloy and process |
| Riser design database | Riser types, dimensions, and feeding capacity data |
| Gating system database | Gate dimensions, choke area calculations, system ratios |
3.3 Temperature Field Simulation Subsystem
This subsystem provides comprehensive functionality for thermal analysis of the casting solidification process. The material property database contains thermophysical properties such as thermal conductivity, density, specific heat, and enthalpy for common casting alloys and mold materials. The finite element mesh generation module creates appropriate mesh density distributions based on geometric complexity and analysis accuracy requirements. The solution module handles initial conditions, boundary conditions, and iterative convergence control. Result processing includes color contour plots, temperature history curves, and dynamic animation of solidification progress.
3.4 Defect Prediction Subsystem
The defect prediction subsystem employs two primary criteria: the temperature gradient method and the Niyama criterion method. Both approaches analyze temperature field simulation results to identify regions susceptible to shrinkage cavity and porosity formation.
Temperature Gradient Method
Extensive practical experience demonstrates that shrinkage cavities in casting centerlines are governed by temperature gradients. When temperature gradients are large, the expansion angle of the feeding channel toward the riser is wide, enabling effective riser feeding and preventing defect formation. Conversely, small temperature gradients result in premature closing of feeding channels. The critical temperature gradient value varies with casting shape, dimensions, and alloy composition, making precise determination challenging.
Niyama Criterion Method
Developed by Niyama, this criterion predicts shrinkage porosity based on the ratio of temperature gradient to cooling rate. The Niyama value (G/√R) is calculated for each element, and when this value falls below a critical threshold, the element is considered a potential defect location. According to Niyama, the critical value is 1.0 °C·s1/2/cm3/2, while Li Wenzhen’s research suggests the critical value depends on casting size, ranging from 0.8 to 1.5 °C·s1/2/cm3/2.
4. Development Technologies and Implementation Approach
4.1 Expert System Principles and Techniques
The defect analysis subsystem is developed based on expert system technology, which utilizes knowledge representation, inference mechanisms, and explanation facilities to simulate human expert decision-making processes. An expert system consists of six primary components: knowledge base, working database, inference engine, knowledge acquisition mechanism, explanation mechanism, and user interface.
Knowledge representation in this system uses production rules of the form:
$$IF\ (conditions)\ THEN\ (conclusions)\ WITH\ (certainty\ factor)$$
For example, a production rule for identifying shrinkage cavities might be expressed as:
IF the defect is a hole-type defect AND the shape is irregular enclosed or exposed AND the hole wall is rough with dendritic crystals AND the location is at hot spots or final solidification regions AND the holes are large and concentrated THEN the defect is classified as shrinkage cavity (confidence factor = 0.95)
The inference mechanism combines forward and backward chaining strategies. Forward chaining selects initial targets, while backward chaining solves specific goals, improving efficiency and saving memory space. To handle uncertainty, the system employs certainty factors with fuzzy mathematics and probability theory.
4.2 Internal Development Tools
ANSYS provides three development tools for customization: the ANSYS Parametric Design Language (APDL) and User Interface Design Language (UIDL) are the most extensively utilized in this system. APDL enables macro definition, toolbars creation, parameter passing, mathematical operations, and database management. UIDL allows modification and creation of menus, dialogs, and help files.
Table 4 summarizes the application of development tools within the system:
| Tool | Application Areas |
|---|---|
| APDL | Parametric modeling, mesh generation, post-processing, data extraction, automation of simulation workflows |
| UIDL | Custom menus, input dialogs, help system, knowledge-base interface |
| Visual Basic | External database applications, user interface enhancement, knowledge management tools |
| Microsoft Access | Background database management for material properties, knowledge base, and defect information |
4.3 Three-Layer Development Architecture
The system architecture follows a three-layer hierarchical structure:
- Application layer: The casting defect prediction and analysis system itself
- Tool layer: Internal ANSYS databases, external development tools, and knowledge bases
- System layer: The underlying operating system environment
This architecture provides excellent compatibility, extensibility, and development efficiency. The modular approach allows flexible updates and enhancements to individual components without affecting the entire system.
5. Shrinkage Cavity and Porosity Prediction Methodology
5.1 Classification and Formation Mechanisms
Shrinkage cavities are classified based on their formation mechanisms: hot spot shrinkage, riser neck shrinkage, ingate shrinkage, axial dispersed shrinkage, re-entrant corner shrinkage, core-induced shrinkage, and mold displacement shrinkage. Porosity is similarly categorized into interdendritic porosity, hot spot porosity, axial porosity, and mold displacement porosity.
The fundamental mechanisms governing shrinkage cavity formation include:
- Metal flow and pressure drop: Liquid contraction, solidification contraction, and mold deformation cause liquid level depression and pressure reduction during solidification
- Solute concentration and reactions: Gas solubility decreases with temperature decline, promoting bubble formation
- Interdendritic feeding failure: When solid fraction exceeds the critical value, liquid flow becomes impossible through the dendritic network
The Darcy’s law describes flow through the mushy zone:
$$v = -\frac{K}{\mu} \left(\frac{dp}{dx} – \rho g\right)$$
where K is permeability, μ is dynamic viscosity, dp/dx is pressure gradient, ρ is liquid density, and g is gravitational acceleration. The pressure drop due to flow resistance can be expressed as:
$$\Delta p = \frac{\mu(1-f_s)}{K} \cdot \Delta x \cdot v$$
As the solid fraction increases, pressure losses increase dramatically, eventually preventing liquid flow and causing shrinkage defects.
5.2 Temperature Gradient Method Implementation
The temperature gradient method is implemented within the system using the following computational approach. First, the solid fraction for each element is calculated using either linear or parabolic distribution models based on user-input liquidus and solidus temperatures. The system includes automatic validation to prevent incorrect temperature input (e.g., liquidus temperature lower than solidus temperature triggers an error message).
The user interface for critical parameter input includes:
- Liquidus and solidus temperatures
- Critical solid fraction
- Critical temperature gradient value
- Analysis range selection (whole simulation or specific load steps)
The system workflow allows users to either analyze the entire simulation process or focus on specific time steps. The “smart analysis” option automatically processes all load steps, while “load step analysis” enables targeted investigation of particular moments. The input is validated and errors are reported through user-friendly dialog messages.
5.3 Niyama Criterion Method Implementation
The Niyama criterion method requires additional input of initial temperature and critical Niyama value. The system calculates the Niyama value for each element:
$$G/\sqrt{R}$$
where G is the temperature gradient and R is the cooling rate. Elements with values below the critical threshold are identified as potential defect locations. The workflow follows a similar pattern to the temperature gradient method, allowing both whole-simulation and load-step-specific analysis.
Table 5 summarizes the input parameters required for both prediction methods:
| Parameter | Temperature Gradient Method | Niyama Criterion |
|---|---|---|
| Liquidus temperature | Yes | Yes |
| Solidus temperature | Yes | Yes |
| Critical solid fraction | Yes | Yes |
| Critical temperature gradient | Yes | No |
| Critical Niyama value | No | Yes |
| Initial temperature | No | Yes |
6. Defect Analysis Expert System Development
6.1 Defect Classification Systems
Proper identification of casting defects is essential for effective defect analysis. This system integrates multiple classification approaches to provide comprehensive analysis capabilities. The primary classification is based on the International Committee of Foundry Technical Associations (ICFTA) system, which organizes defects into seven categories with four-level hierarchical coding (one letter plus three digits). Table 6 presents the ICFTA classification categories:
| Code | Category | Number of Defect Types |
|---|---|---|
| A | Metallic projections | 17 |
| B | Cavities | 12 |
| C | Cracks | 18 |
| D | Surface defects | 22 |
| E | Incomplete castings | 8 |
| F | Incorrect dimensions or shape | 10 |
| G | Inclusions or metallurgical anomalies | 12 |
Additionally, this system proposes a physical field-based classification that connects defects to the simulation physical fields that predict them:
| Physical Field | Defect Types |
|---|---|
| Flow field | Cold shut, gas entrapment, slag inclusion, mold erosion |
| Temperature field | Shrinkage cavity, shrinkage porosity |
| Stress-strain field | Hot tearing, cold cracking, deformation |
| Flow-thermal coupling | Misrun, cold shut |
6.2 Knowledge Base Design
The knowledge base constitutes the core of the defect analysis subsystem. Its structure includes defect information tables, classification tables, cause tables, and prevention measure tables as shown in Table 8.
| Table Name | Contents |
|---|---|
| Defect number table | Unique identifier for each defect |
| Chinese name table | Chinese terminology for each defect |
| English name table | English terminology for each defect |
| Defect category table | Classification hierarchy information |
| Classification system table | Mapping between different classification systems |
| Defect cause table | Potential causes for each defect type |
| Defect countermeasure table | Preventive measures and recommendations |
6.3 Inference Rules and Reasoning Process
The inference engine employs backward chaining to identify defect attributes based on user selections. The reasoning process for identifying shrinkage cavity may proceed as follows:
- Is the defect of the cavity type? If yes, continue
- Is the cavity shape irregular, closed or exposed? If yes, continue
- Is the cavity wall rough with dendritic features? If yes, continue
- Is the cavity located at hot spots or final solidification regions? If yes, continue
- Are the cavities large and concentrated? If yes, conclude shrinkage cavity
The system considers production process factors that affect casting quality, categorized into 10 groups shown in Table 9, plus simulation analysis factors shown in Table 10.
| Category | Process Elements |
|---|---|
| 1 | Casting design and pattern design |
| 2 | Pattern equipment preparation |
| 3 | Flask and molding box preparation |
| 4 | Gating and risering system |
| 5 | Mold sand composition |
| 6 | Core making |
| 7 | Molding operation |
| 8 | Metal composition and melting |
| 9 | Pouring practice |
| 10 | Other factors |
| Factor | Description |
|---|---|
| Mesh density | Spatial discretization affects solution accuracy |
| Load application | Initial and boundary condition specification |
| Linear vs. nonlinear analysis | Material property and radiation considerations |
| Unit system consistency | Proper dimension management |
| Load step configuration | Temporal discretization for transient analysis |
6.4 System Application and Maintenance
The defect analysis system provides user-friendly interfaces for selecting defect types, choosing relevant process factors, and displaying countermeasures. The application sequence involves:
- Selecting the defect class and specific defect type from hierarchical menus
- Viewing the international defect code and description
- Selecting casting type and process parameters
- Identifying potential contributing factors from process and simulation categories
- Generating recommendations and preventive measures
Knowledge base maintenance functionality allows authorized users to modify existing knowledge entries, add new knowledge, and refine inference rules. This capability ensures the system evolves with accumulating production experience and research findings.
7. Case Study: High Chromium Cast Iron Casting
7.1 Casting Description and Process Design
The case study involves a high chromium cast iron hammer head for a crusher, operating under impact conditions requiring high abrasion resistance and adequate toughness. The rectangular casting contains a central cavity, representing a moderately complex geometry. The chemical composition of the alloy is presented in Table 11.
| Element | Content (wt%) |
|---|---|
| C | 2.4-2.8 |
| Si | 0.3-0.8 |
| Mn | 0.4-0.8 |
| Cr | 18-22 |
| Mo | 0.5-1.0 |
| Cu | 0.5-1.0 |
| Ni | 0.3-0.8 |
| P | ≤ 0.06 |
| S | ≤ 0.05 |
The original process employs a gating system that also serves as a riser, with four chill blocks placed in the lower mold and a weighted sand core in the center. Three-box molding is used with green sand as the mold material, resin sand for cores and weights, and carbon steel for chills.
7.2 Solidification Characteristics
Experimental temperature measurements identified two distinct solidification stages. In the first stage (before approximately 1000 seconds), all thermocouple points remain above the liquidus temperature of approximately 1260°C, with the riser region cooling slowest and the weighted region cooling fastest. The maximum measured temperature was approximately 1250°C. During this period, austenite dendrites nucleate and grow from the liquid phase with primary austenite formation.
In the second stage (after 1000 seconds), eutectic reaction begins at the casting bottom, releasing latent heat and altering cooling rates. The eutectic reaction temperature range is approximately 1150-1200°C. The solidification interval is relatively wide, favoring volume solidification behavior. Austenite dendrites continuously grow, dividing remaining liquid into small melt pools that readily form dispersed porosity.
7.3 Temperature Field Simulation
Thermophysical properties for the simulation are established from material databases and experimental data. Table 12 presents the key thermal property values used in the simulation.
| Material | Property | Value | Unit |
|---|---|---|---|
| Casting (High Cr iron) | Density | 7500 | kg/m³ |
| Specific heat | 460-700 | J/(kg·°C) | |
| Thermal conductivity | 20-35 | W/(m·°C) | |
| Enthalpy | Loaded from database | J/m³ | |
| Chill (Medium carbon steel) | Density | 7850 | kg/m³ |
| Specific heat | 434-640 | J/(kg·°C) | |
| Thermal conductivity | 30-50 | W/(m·°C) | |
| Mold (Green sand) | Density | 1550 | kg/m³ |
| Specific heat | 840-1090 | J/(kg·°C) | |
| Thermal conductivity | 0.6-1.2 | W/(m·°C) | |
| Core & weight (Resin sand) | Density | 1600 | kg/m³ |
| Specific heat | 800-1050 | J/(kg·°C) | |
| Thermal conductivity | 0.6-1.0 | W/(m·°C) |
The initial temperature condition is established at 1200°C for casting and gating system components, and 50°C for mold assemblies. Boundary conditions assume perfect contact between casting and mold (conduction only), while external surfaces transfer heat through combined convection and radiation, with radiation effects incorporated into adjusted convection coefficients.
Simulation results reveal localized high-temperature regions and low-temperature-gradient zones at the inner cavity corners, where feeding channels are constricted despite the riser and gating system maintaining elevated temperatures. These locations correspond closely with actual defect occurrences reported in production statistics.
7.4 Defect Prediction Results
To determine critical threshold values for the prediction criteria, two approaches are proposed: the forward method (adjusting initial critical values through iterative simulation-experiment comparison) and the reverse method (starting from optimized processes). This study adopts the forward method for the high chromium cast iron casting.
Temperature Gradient Method Results
Through multiple simulation iterations and comparison with experimental observations, the critical temperature gradient value is determined to be 1.5°C/mm for this high chromium cast iron casting. The temperature gradient method successfully predicts shrinkage cavity formation in the gating system, secondary shrinkage below primary cavities, and porosity in thick sections, all consistent with production statistics.
Niyama Criterion Results
In contrast, the Niyama criterion predictions deviate significantly from actual observations. This discrepancy suggests that criteria successful for steel castings may not transfer directly to cast iron applications. The solidification behavior of cast iron involves simultaneous austenite contraction and graphite expansion, creating complex volumetric changes. Therefore, the Niyama criterion proves unsuitable for predicting sand foundry defects in high chromium cast iron components.
7.5 Process Optimization
Based on identified limitations of the original process, optimization measures include adding padding adjacent to the riser walls and replacing the rectangular riser with a square cross-section riser of equal volume. The square riser provides a larger modulus for the same volume, extending solidification time and improving feeding efficiency.
Table 13 compares key parameters between the original and optimized processes:
| Parameter | Original Process | Optimized Process |
|---|---|---|
| Riser dimensions (mm) | Rectangular 150×120×100 | Square 125×125×115 |
| Riser modulus | Approximately 23 | Approximately 32 |
| Padding | None | Added (12 mm thickness) |
| Feeding efficiency | Limited | Improved |
Simulation of the optimized process demonstrates that riser and gating regions retain elevated temperatures at the completion of casting solidification, enabling more complete feeding of potential shrinkage regions. Defect prediction results for the optimized process indicate no defects within the casting itself, confirming the effectiveness of the process modifications.
8. Conclusions and Future Prospects
8.1 Key Achievements
This research successfully develops a casting defect prediction and analysis system based on the ANSYS platform. The principal achievements include:
- Comprehensive system architecture: The system incorporates casting process CAD, temperature field simulation, defect prediction, and defect analysis capabilities, with seamless information flow between subsystems.
- Improved defect prediction algorithms: The linear solid fraction distribution method is successfully applied to sand foundry defect prediction. Both the temperature gradient and Niyama criteria are implemented with flexible analysis range options (whole simulation or specific load steps).
- Physical field-based defect classification: A novel classification system organizes defects according to the physical fields responsible for their formation, facilitating integration of simulation results with defect analysis.
- Expert system-based defect analysis: The defect analysis subsystem provides defect inquiry, cause analysis, prevention recommendations, and knowledge base maintenance functionality through integrated rule-based reasoning.
- Achievement of critical value determination: The research proposes forward and reverse methods for determining critical criteria values. Through forward analysis, the critical temperature gradient for high chromium cast iron is established at 1.5°C/mm.
8.2 Validation Results
Application to a high chromium cast iron hammer head demonstrates that the temperature gradient method accurately predicts shrinkage cavity and porosity locations matching production observations. The Niyama criterion proves unsuitable for this material system due to the unique contraction-expansion behavior of cast iron during solidification. Process optimization guided by defect prediction successfully eliminates predicted defects, confirming the practical value of the developed system.
8.3 Limitations and Future Work
Several limitations require further development:
- Additional prediction criteria: Incorporating other defect prediction methods (e.g., pressure gradient method, solid fraction gradient method, direct simulation approach) would provide multiple analysis perspectives.
- Quantitative defect grading: Improved methods for quantitatively grading shrinkage defect severity based on quality requirements.
- Enhanced expert system: Further development of inference rules, knowledge acquisition mechanisms, and self-learning capabilities.
- Extended user interface: More comprehensive parameter input regions, schematic displays, and online assistance to simplify the analysis process.
The future development of this system should continue to focus on practical engineering applications, with validation cases derived from real production scenarios. The goal remains to facilitate the application of casting simulation and expert system technologies to improve casting quality, reduce development cycles, and reduce production costs in the foundry industry. The advancement from empirical knowledge toward scientific analysis, from qualitative descriptions to quantitative predictions, marks an essential evolution for sand foundry defect prevention and overall casting quality enhancement.