AI helps find better recipes for 3D-printing titanium parts

Istanbul, Türkiye - Engineers can 3D-print high performance titanium parts for aircraft, medical implants, and advanced machinery, but finding the right printing settings can still require costly and time-consuming trial and error. A new study shows how artificial intelligence could help narrow that search before production testing begins. 

Researchers have developed an explainable machine-learning framework for Selective Laser Melting, a metal 3D-printing process that melts titanium alloy powder layer by layer with a laser. The work focuses on Ti6Al4V, a widely used titanium alloy valued for its high strength-to-weight ratio, corrosion resistance, and biocompatibility. 

The collaboration brings together Istanbul Technical University and Gazi University in Türkiye with partners from the University of Engineering and Technology, Lahore; Pak-Austria Fachhochschule: Institute of Applied Sciences and Technology; the University of Pittsburgh; and the University of Udine. 

The study, published in the International Journal of Precision Engineering and Manufacturing, addresses a practical challenge in metal additive manufacturing: small changes in laser power, scanning speed, layer thickness, hatch spacing, laser spot size, and powder size can determine whether a printed part is strong, ductile, or inconsistent in performance. 

To build the framework, the research team compiled 201 experimental records from published studies. These records linked six SLM process parameters to three key tensile properties: yield strength, ultimate tensile strength, and elongation at fracture. The team compared several machine-learning approaches and found that an artificial neural network model produced the most consistent predictions across the three mechanical properties. The researchers then applied SHAP analysis, an explainable AI method, to show how different printing parameters contributed to the model’s predictions. 

The most practical feature of the study is a reverse-design graphical user interface. Unlike models that only predict material properties from a given set of printing parameters, this tool allows users to enter the desired strength and ductility and then receive suggested process settings for further testing. By narrowing the range of promising settings before experiments begin, the approach could reduce unnecessary trial-and-error, saving development time, material, and cost.

“Metal additive manufacturing still requires careful process control, especially for demanding applications such as aerospace structures and biomedical implants,” said Prof. Hüseyin Kızıl of Istanbul Technical University, who supervised the work. “Our aim is to provide engineers with a data-driven decision tool to narrow the process window before experimental validation begins.” 

The researchers emphasize that the tool is not intended to replace physical testing, quality assurance, or certification. Instead, it can support pre-production evaluation by helping engineers focus experiments on more promising parameter combinations and by improving understanding of the relationship between printing conditions and material performance. The approach may be useful for companies and research groups developing titanium components where reliability, precision, and material efficiency are critical. With further validation using machine-specific datasets and under real production conditions, AI-guided process design could help shorten development cycles, reduce material waste, and support more consistent manufacturing of high-performance metal parts. 

The team welcomes collaboration with industrial partners and research groups interested in validating the framework with production-relevant SLM datasets, machine-specific parameters, and application-specific performance requirements.