Secure your spot now for exclusive AM insights at our Additive Manufacturing Advantage online conference covering aerospace, space, and defense!
Faculty members from Arizona State University‘s (ASU) School of Computing and Augmented Intelligence are working to solve a long-standing challenge in metal 3D printing: how to reliably predict the internal structure and strength of printed components.
Their effort combines computer science, manufacturing, and artificial intelligence (AI) to improve how stainless steel parts are formed, with the aim of bringing greater consistency and precision to the AM process. Part of the Ira A. Fulton Schools of Engineering, the project is called “CompAM: Enabling Computational Additive Manufacturing,” funded by the National Science Foundation (NSF).
It is being led by Computer Science (CS) Professor Aviral Shrivastava and Industrial Engineering Professor Ashif Iquebal. The core idea is to develop an AI-driven system that can forecast how a metal’s microstructure will evolve during printing, something that typically requires intensive simulations or costly trial-and-error approaches.
“In industries like aerospace, defense and energy, the performance of metal components isn’t negotiable — it’s mission-critical. By giving manufacturers faster, more accurate tools to predict and control material properties, we’re enabling a new era of precision manufacturing and reducing the costly guesswork that often slows innovation,” says Iquebal.
AI-guided predictions in additive manufacturing
A key challenge in metal AM is that cooling conditions during printing alter the material’s microstructure, which in turn affects its mechanical properties. Even minor thermal variations can lead to significant performance differences. Engineers have traditionally addressed this through complex simulations or physical testing, both of which are time-consuming and computationally expensive.
With their project, the researchers at ASU are proposing a more efficient path. Their AI model blends established physics equations with data-driven learning to make faster, more targeted predictions about material behavior.
Rather than analyzing the entire object in full detail, the system identifies which regions of a part are most sensitive to thermal changes and allocates computing power there. This approach reduces the time and cost of simulations while preserving their predictive accuracy.
To demonstrate their approach, the team is producing a naval propeller from 316L stainless steel, chosen for its industrial relevance and challenging geometry. They aim to control the metal’s grain size to under 1 µm to enhance durability and performance.
The project also reflects a deliberate effort to make advanced manufacturing tools more useful in real-world industrial settings. The researchers are using a 3D printer at ASU’s Innovation Hub, equipped with a six-axis robotic arm and high-power lasers, to fabricate the propeller. Once printed, the part’s actual microstructure will be compared to what the AI predicted, and also benchmarked against results from traditional simulations.
If the method proves accurate, manufacturers could soon have a faster way to fine-tune printing conditions to get exactly the material properties they need without repeating long testing cycles.
The team plans to release their software and tools as open-source resources, allowing researchers and industry professionals to adapt them for their own use. The project will also be incorporated into graduate-level computer science education and K–12 STEM outreach.
Predicting metal behavior during additive manufacturing
ASU’s effort joins a growing body of work focused on improving prediction and control in metal AM.
At the 2025 AMUG Conference, Flow Science’s computational fluid dynamics (CFD) engineer Garrett Clyma showcased how high-fidelity simulation is advancing metal AM by modeling melt pool behavior to reduce defects and improve process control. Using FLOW-3D AM, engineers can simulate complex thermal interactions, assess beam shaping strategies, and visualize melt pool stability without costly experimentation.
Studies validated with in-situ X-ray imaging showed that ring and spiral laser beam profiles produced more stable melt pools compared to traditional Gaussian beams. The software captures critical factors like surface tension, absorptivity, and cooling rates, allowing users to infer microstructural outcomes. With expanded simulation capabilities, engineers can now evaluate trade-offs and optimize printing conditions before starting a build.
In 2023, researchers from the National Institute of Standards and Technology (NIST), KTH Royal Institute of Technology confirmed how cooling rates during laser powder bed fusion (LPBF) influence the crystal structure of 3D printed metals. Using synchrotron x-ray imaging at Argonne and the Paul Scherrer Institute, the team observed solidification in hot-work steel at cooling rates up to 1.5 million kelvins per second.
Higher cooling rates suppressed ferrite, which is associated with cracking, and promoted the formation of austenite, a more desirable structure. These findings validate the Kurz-Giovanola-Trivedi (KGT) model, allowing manufacturers to predict and control microstructure during printing. Published in Acta Materialia, the study supports more consistent and scalable metal AM through simulation-based approaches.
What 3D printing trends should you watch out for in 2025?
How is the future of 3D printing shaping up?
To stay up to date with the latest 3D printing news, don’t forget to subscribe to the 3D Printing Industry newsletter or follow us on Twitter, or like our page on Facebook.
While you’re here, why not subscribe to our Youtube channel? Featuring discussion, debriefs, video shorts, and webinar replays.
Featured image shows an illustration of the metal 3D printing process. Image via Andrea Heser/ASU.
