AI4MAT - AI Driven Material Selection for Machine Tool Design
Accelerating engineering workflows through data driven material optimisation
Ce.S.I. (Centro Studi Industriali) is an Italian SME specialising in advanced engineering services, including mechanical design, multiphysics simulation, digital twins, machine learning, AI and manufacturing support. The company supports industrial customers in developing innovative production equipment through a combination of deep technical expertise and cutting edge digital capabilities.
However, traditional physics based simulations (white box models) used for material selection in machine tool design had become a significant bottleneck. These models required extensive computation time, suffered from missing knowledge and uncertainties, and limited Ce.S.I.'s ability to efficiently evaluate alternative material behaviours during early design phases. To overcome these limitations, Ce.S.I. joined the AI4MAT pilot within the AIRISE project, aiming to integrate AI driven, data based modelling techniques to accelerate early stage material selection and shorten the development cycle.
Traditional Dairy Production
AI-Powered Engineering Optimisation for Material Selection
- Limitations of conventional simulation models:
- Existing white box physics models lacked completeness, introduced uncertainty and required long computational loops.
- Time consuming design cycles:
- Material selection depended heavily on iterative FEM/CAE simulations, making early stage engineering slow and resource intensive.
- Complex multi criteria decision making:
- Engineers needed to balance mass, stiffness, material type and cost - often requiring repeated simulation-based validation.
- Adoption ready reliability and transparency:
- The AI model needed not only high predictive performance, but also explainability to ensure engineers could trust and validate its output.
The AI4MAT pilot was conducted over a defined time horizon and included controlled engineering scenarios representative of actual machine tool design conditions.
Ce.S.I. (Centro Studi Industriali) and the AIRISE project jointly:
- Developed and benchmarked multiple AI models:
- Including Random Forests, Neural Networks and fusion based learning approaches for material classification.
- Integrated Explainable AI (XAI):
- Using SHAP, SHAP based feature importance and permutation importance to improve model transparency and support engineering interpretation.
- Aligned evaluation to engineering KPIs:
- Focusing on accuracy, robustness, design throughput efficiency and suitability for practical engineering workflows.
- Maintained strong cross team collaboration:
- Through continuous model tuning, iterative validation and technical knowledge exchange to ensure the AI outputs aligned with Ce.S.I.'s industrial constraints.
- Custom AI tool for early stage material selection:
- A tailored AI solution was developed to support design engineers in identifying feasible material candidates before running extensive CAE simulations.
- High predictive accuracy (up to 98%):
- The AI model demonstrated strong reliability for material classification tasks.
- Substantial efficiency gains:
- 97% reduction in time required for initial material screening.
- 20% reduction in the overall product development cycle.
- Improved transparency and trust through XAI:
- Explainability enabled engineers to understand why the model suggested certain materials, supporting internal acceptance and confident adoption.
- Demonstrated industrial impact:
- The pilot validated that AI driven material selection can meaningfully enhance real engineering processes in machine tool design.
- The pilot validated that AI driven material selection can meaningfully enhance real engineering processes in machine tool design.
Ce.S.I. (Centro Studi Industriali) plans to:
- Increase the Technology Readiness Level (TRL) from 6–7 to 9:
- Transitioning from experimental validation to full industrial implementation.
- Expand AI assisted design to additional engineering phases:
- Embedding data driven decision support throughout the entire development pipeline.
- Improve robustness through broader validation:
- By testing across more material types, operating conditions and machine tool configurations.
- Continue collaboration with AI Rise and Chemiq:
- Focusing on further tuning, validation and strengthening the AI models for long‑term integration.