Introduction
Accurate control of process parameters is critical to achieving consistent quality in metal Additive Manufacturing (AM), especially for thermally conductive materials like aluminum. Aluminum AM parameter optimization with AI is at the core of ValCUN’s solution: its Molten Metal Deposition (MMD) process—a novel AM technology inspired by Fused Filament Fabrication (FFF)—builds aluminum parts in a single, streamlined step.
To enhance process control, ValCUN has developed a very specific Artificial Intelligence (AI) model named Artificial Neural Network (ANN). This model predicts optimal printing parameters faster than traditional simulation methods, helping to ensure consistent layer bonding and improve overall manufacturing efficiency.
The Challenge: Thermal Control in Aluminum Additive Manufacturing
In aluminum AM processes like MMD, layer fusion quality is a key factor. Too little energy results in poor bonding; too much can overheat or distort the part. Because aluminum rapidly conducts heat, finding the right energy input per layer is technically demanding.
Traditional FE Modelling: Accurate But Time-Consuming
ValCUN initially used Finite Element (FE) simulations to optimize MMD parameters. While effective, these physics-based models can take several days to deliver results—especially for complex toolpaths or large components—making them impractical for real-time applications.
ValCUN’s Solution: AI-Powered Optimization with ANN
To overcome this bottleneck, ValCUN has trained an Artificial Neural Network (ANN) using FE-generated datasets. Once trained, the ANN can predict the optimal deposition temperature for any toolpath scenario in just seconds—supporting real-time adjustments during the printing process.
Case Study: Benchmark Part Optimization with AI
Part Complexity & Toolpath Sensitivity
The benchmark component features steep overhangs, junctions, and varying wall thicknesses—all factors that significantly affect local fusion energy requirements. This complex geometry was used to test the ANN’s ability to replicate FE simulation results.
Results: ANN vs FE Modelling
- FE Model Runtime: ~8 days (16-core CPU, 64 GB RAM)
- ANN Prediction Time: ~10 seconds
- Accuracy: High agreement between ANN and FE thermal maps (see Figure 3), with minor local variations
Key Benefits of ANN-Driven Process Optimization
- Speed: Real-time parameter prediction for toolpath-sensitive processes
- Scalability: Easily extended to new part geometries and sizes
- Energy Control: More consistent layer fusion in thermally conductive metals
- Efficiency: Reduces trial-and-error calibration time during part development
Applications Beyond MMD
While developed for ValCUN’s MMD process, the ANN framework can be adapted to other metal AM technologies, including Wire Arc Additive Manufacturing (WAAM) and Directed Energy Deposition (DED)—anywhere toolpath-sensitive thermal control is a challenge.
Conclusion
ValCUN’s integration of Aluminum AM parameter optimization with AI into aluminum additive manufacturing marks a significant step forward in predictive process control. By combining physics-based simulations with Artificial Neural Networks (ANN), MMD becomes smarter, faster, and more adaptable—unlocking higher productivity and part quality for aluminum 3D printing. This aligns with industry trends showing that AI methodologies have become integral to metal AM, offering unprecedented capabilities in process optimization, property prediction, and defect detection