Additive manufacturing, or 3D printing, offers the ability to produce complex parts on demand using a variety of materials, from ceramics to nylon. However, predicting the strength and durability of metal parts created through this process remains a significant challenge.
Researchers at Arizona State University’s School of Computing and Augmented Intelligence are developing advanced artificial intelligence tools to enhance the speed and reliability of 3D printing with stainless steel. Professors Aviral Shrivastava and Ashif Iquebal have secured funding from the National Science Foundation for their project “CompAM: Enabling Computational Additive Manufacturing.” Their goal is to use AI to predict how stainless steel's internal structure forms during printing, enabling faster and more precise metal additive manufacturing.
Printing a Naval Propeller: A Practical Test Case
The team’s demonstration involves 3D printing a five-axis naval propeller using 316L stainless steel, a common industrial-grade alloy. Propellers are large, geometrically complex parts requiring exact manufacturing—making them ideal for testing the limits of metal 3D printing.
Controlling Microstructure Through AI
The internal microstructure of a metal part critically influences its performance, and this microstructure depends on details like cooling rate and thermal history during printing. Traditional manufacturing relies on methods such as quenching to control these properties. Fast cooling produces hard, brittle metals, while slower cooling yields more malleable materials.
In metal additive manufacturing, small changes in heat or timing can drastically affect the microstructure and, consequently, the mechanical properties of the final product. Current industry approaches rely on complex simulations or trial and error to optimize these parameters, which can be time-consuming and costly. For example, simulating one set of printing parameters for the naval propeller can take over 60 days on a 1,000-core supercomputer.
Physics-Informed AI to Accelerate Simulations
Shrivastava and Iquebal's team is building a physics-informed AI system that predicts microstructure formation during printing more efficiently. This approach combines physical rules with data-driven learning, allowing the AI to focus on critical data points without requiring massive datasets or computational power.
The system identifies which regions of the metal need detailed attention and which can be approximated, reducing simulation time significantly. This method can help manufacturers fine-tune printing parameters to achieve desired material properties with fewer trial runs.
Bridging Research and Industry
Iquebal emphasizes the project's industrial relevance, especially for sectors like aerospace, defense, and energy where component performance is critical. Providing faster and more accurate predictive tools helps reduce costly guesswork and accelerates innovation in precision manufacturing.
The team uses a six-axis robotic arm-equipped 3D printer at the ASU Innovation Hub to produce the naval propeller. They plan to compare the AI-predicted microstructure with actual measurements from the printed part and benchmark these results against traditional simulations.
By making their software and tools freely available, the researchers aim to support broader adoption and advancement in additive manufacturing simulation capabilities.
Educational and Outreach Impact
Beyond research, the project will be integrated into graduate-level computer science courses and outreach programs for K–12 students. This initiative highlights the intersection of industrial engineering and computer science and demonstrates how AI can enhance materials science and manufacturing processes.
For professionals interested in expanding their AI skills, exploring courses on Complete AI Training can provide practical knowledge applicable to manufacturing and engineering challenges.
Your membership also unlocks:
