Abstract
The optimization of investment casting processes heavily relies on empirical trial-and-error methods, resulting in long optimization cycles, high labor costs, low operational efficiency, and a lack of algorithmic optimization. Therefore, an integrated computing platform for investment casting processes was developed based on the ProCAST finite element simulation software and multitasking batch processing technology. This platform integrates various functions, including Design of Experiments (DOE) algorithms, finite element simulation, automatic extraction of result data, surrogate model construction, and collaborative multi-objective optimization. Taking a turbine guide vane as an example, this paper realizes integrated optimization of casting process-defect simulation with shrinkage as the target value and pouring temperature, ambient temperature, and thermal emissivity as variables. Compared with traditional simulation optimization, the efficiency of the integrated computing platform is improved by 91.66%.
1. Introduction
The research and development mode of high-temperature alloy castings for aero-engine in China faces many challenges in achieving high quality and efficiency. Even with known material compositions, it is still difficult to manufacture components with completely acceptable dimensional accuracy. The traditional “empirical optimization” approach, characterized by experience accumulation and simple cyclical trial-and-error, cannot meet the precise and reliable requirements for process optimization and decision support. Therefore, the development of digital and intelligent casting theoretical methods and technologies is urgent.
Table 1. Research Background and Challenges
| Research Area | Challenge |
|---|---|
| High-Temp Alloy Castings | Difficulty in achieving dimensional accuracy |
| Optimization Method | Empirical trial-and-error lacks precision and reliability |
2. Integrated Computational Platform for Investment Casting
2.1 Platform Design
The design of the integrated computational platform for investment casting mainly consists of four parts:
- Automation: Reducing human operation through algorithmic implementation of tasks such as process parameter optimization, simulation, and result analysis.
- Multi-Objective Optimization: Integrating DOE algorithms (e.g., Latin Hypercube, Box-Behnken, Central-Composite) and optimization algorithms (e.g., NSGA-Ⅱ, NLPQLP, Multi-Island GA) for multi-factor optimization.
- Process Flow Management: Covering multiple stages from process parameter setting to result validation and optimization.
- Cross-Platform Usability: Operating on both Windows Server and Linux systems based on computational needs.
Table 2. Platform Design Components
| Component | Description |
|---|---|
| Automation | Reducing human operation through algorithms |
| Optimization | Integrating DOE and optimization algorithms |
| Process Flow | From parameter setting to result validation |
| Cross-Platform | Usable on Windows Server and Linux |
2.2 Platform Architecture
The integrated computational platform for investment casting is developed using Python, with a B/S (Browser/Server) framework separating the frontend and backend. The frontend uses Flask, and the backend uses Django, making it accessible through any browser. The platform integrates ProCAST underlying scripts and cooperates with high-performance computing servers for high-throughput calculations.
2.3 Platform Composition
The platform is divided into three major modules: integrated computing, integrated results, and job management.
Table 3. Platform Modules and Functions
| Module | Function |
|---|---|
| Integrated Computing | Selects process factors, integrates DOE algorithms, and generates ProCAST scripts for parallel simulation |
| Integrated Results | Processes simulation results, extracts porosity volumes, builds approximate models, and screens optimal solutions |
| Job Management | Schedules jobs, submits tasks to high-performance servers, and monitors progress in real-time |
3. Application Example: Turbine Guide Vane
The turbine guide vane is a critical component in turbines, used to guide and control gases or fluids in the working flow, optimizing turbine performance and efficiency.
Table 4. Turbine Guide Vane Specifications
| Specification | Value |
|---|---|
| Material | K418B |
| Max. Diameter | ϕ527 mm |
| Height | 94 mm |
| Number of Blades | 58 |
| Blade Exhaust Edge Length | 100 mm |
| Exhaust Edge Radius (R) | 0.35 mm |
| Shell Material | Refractory Mullite |
| Shell Thickness | 6 mm |
3.1 Integrated Calculation Process
- Mesh Generation: Imports the 3D model, generates mesh files using ProCAST scripts.
- Boundary Condition Setting: Sets parameters such as pouring temperature, convective heat transfer coefficient, thermal emissivity, and ambient temperature.
- Simulation Execution: Executes simulations in parallel on high-performance servers.
- Result Extraction: Automatically extracts porosity volumes using generated scripts.
- Optimization: Builds an approximate model and uses optimization algorithms to find the optimal solution.
3.2 Integrated Result Processing
The platform uses multithreading and distributed computing to process simulation results in parallel. After all jobs are completed, the platform automatically extracts and processes STL files containing porosity models, quickly obtaining porosity volumes.
Table 5. Experimental Design Variables and Their Ranges
| Variable | Upper Limit | Lower Limit |
|---|---|---|
| Pouring Temperature (°C) | 1550 | 1450 |
| Thermal Emissivity | 0.7 | 0.5 |
| Ambient Temperature (°C) | 100 | 50 |
3.3 Approximate Model and Optimization
An approximate model is built based on experimental data:
Equation: Y = 20.02 – 0.4879×1 + 2.07×2 + 0.7023×3 + 0.1656x1x2 – 1.92x1x3 + 0.5461x2x3 – 4.44×1² + 0.476×2² + 10.2×3²
Where Y is porosity volume, x1 is pouring temperature, x2 is ambient temperature, and x3 is thermal emissivity.
The coefficient of determination (R²) is 0.7693, and the P-value is 0.0032, indicating a good model fit for predictions. The Multi-Island GA algorithm is used to optimize the casting process parameters, resulting in optimal values: pouring temperature of 1549.5 °C, ambient temperature of 50.45 °C, and thermal emissivity of 0.686.
4. Production Verification
The optimized solution obtained from the integrated computing platform was verified in factory production. The turbine guide vane was inspected using X-rays and found to meet the HB20040-2011 standard for aviation high-temperature alloy investment castings, Grade B, and is now produced in batches.
5. Conclusion
- The integrated computing platform successfully integrates DOE, finite element simulation, multi-objective optimization, approximate model construction, and automatic result extraction, enabling collaborative work and automatic optimization of the best simulation scheme.
- Taking the turbine guide vane as an example, an approximate model was built linking pouring temperature, ambient temperature, thermal emissivity, and porosity volume. The optimal process parameters were obtained using the Multi-Island GA algorithm.
- The application of the integrated computing platform significantly improved simulation efficiency, with a 91.66% reduction in time compared to traditional manual finite element simulations.