A team led by Auburn University has secured federal funding to use artificial intelligence to predict how advanced porous materials will behave under extreme conditions, eliminating the need to physically test millions of design variations. The project, part of the White House's Genesis Mission, represents a shift toward AI-assisted materials discovery that combines physics-based modeling with machine learning to accelerate engineering breakthroughs.

What Are These Porous Structures, and Why Do Engineers Care?

The materials at the center of this research are called triply periodic minimal surfaces, or TPMS structures. These are three-dimensional lattices with repeating internal surfaces that create a honeycomb-like pattern. Engineers can manufacture them using 3D printing technology to create parts that are simultaneously lightweight, strong, and capable of managing heat transfer or fluid flow.

The appeal is straightforward: industries from aerospace to biomedical implants need components that weigh less without sacrificing strength or reliability. A solid bone implant, for example, can actually weaken surrounding bone tissue over time because the implant carries all the load. A porous TPMS structure, by contrast, can be engineered to match the strength of natural bone, allowing the patient's skeleton to remain engaged and healthy.

Why Can't Engineers Just Test Everything Physically?

The challenge is scale. The size of the pores, the thickness of the walls, the aspect ratio of each opening, and countless other variables all affect how a structure will perform under fatigue and stress. In theory, engineers could test millions of variations to find the optimal design for each application. In practice, that is impossible.

"The size of the porous openings, the aspect ratio, the thickness of each of those walls, they all matter. They strongly matter," said Reza Molaei, an assistant professor of mechanical engineering at Auburn University. "What I have been doing with my team over the past few years is designing these structures, building them and doing fatigue testing and physics-based fatigue life predictions. Our model has shown very promising predictive capability, but the problem is something else."

Reza Molaei, Assistant Professor of Mechanical Engineering, Auburn University

That problem is the sheer number of untested scenarios. Molaei's team has developed physics-based models that can predict fatigue life with promising accuracy, but those models only cover the specific conditions they have tested. To cover the full range of possible designs and real-world conditions would require testing that is simply not feasible.

How Does AI Solve the Testing Problem?

This is where artificial intelligence enters the picture. Rather than replacing physics with AI, Molaei's approach builds on top of existing physics-based models and uses machine learning to extrapolate predictions across design spaces that have never been experimentally tested. The AI learns the underlying patterns from the tested scenarios and applies those patterns to new conditions.

"We are not replacing physics with AI," Molaei explained. "We are building on physics-based models and using AI to predict the conditions that we are not testing."

Reza Molaei, Assistant Professor of Mechanical Engineering, Auburn University

The Genesis Mission project, announced through Executive Order 14363 signed by President Trump in November 2025, aims to "create AI agents to test new hypotheses, automate research workflow and accelerate scientific breakthroughs" across multiple domains including manufacturing, biotechnology, critical materials, nuclear energy, and quantum information science.

What Real-World Applications Could This Enable?

The potential uses for AI-optimized TPMS structures span multiple industries and address some of engineering's most persistent challenges:

• Aerospace and Defense: Lightweight structures for unmanned aerial vehicles (UAVs) and airplane components that maintain strength while reducing weight, directly improving fuel efficiency and payload capacity.
• Heat Management: Porous heat exchangers and radiators that transfer heat more effectively by increasing surface area, allowing aircraft engines to cool more efficiently without adding weight or bulk.
• Biomedical Implants: Bone implants and other surgical devices engineered to match the strength of natural tissue, preventing bone degradation and improving long-term patient outcomes.
• Structural Resilience: Engineering components designed to absorb energy and resist failure under extreme loads, improving safety and durability in high-stress applications.

How to Accelerate Materials Discovery With AI and Physics

Molaei's team is implementing a structured approach that combines expertise across multiple institutions and disciplines:

• AI Development: Researchers at Kansas State University and Iowa State University are building the machine learning models that will predict material behavior across untested design scenarios.
• Manufacturing and Materials: Oak Ridge National Laboratory is handling additive manufacturing and determining which materials will be used in the TPMS structures, ensuring the designs are practical to produce.
• Mechanical Testing and Modeling: Molaei and Robert Jackson, a mechanical engineering professor at Auburn, are leading the experimental validation and physics-based modeling that forms the foundation for the AI predictions.

This collaborative structure reflects a broader trend in materials science: the most promising breakthroughs require teams that blend computational expertise, experimental rigor, and manufacturing knowledge.

How Competitive Was the Genesis Mission Selection Process?

Molaei's project was among 278 selected for the first phase of the Genesis Mission. The Department of Energy received more than 5,000 project submissions, meaning fewer than 6 percent of proposals made the cut. The selectivity underscores both the ambition of the initiative and the strength of the research community's response.

"Although this is the very first Genesis Mission call, it could be one of my biggest career accomplishments thus far," said Molaei. "I am proud that Auburn is leading this collaboration and grateful for the opportunity to work with an outstanding team of collaborators. I hope that the Phase I effort will establish the scientific and technical foundation for a much larger Phase II project."

Reza Molaei, Assistant Professor of Mechanical Engineering, Auburn University

The Phase I project will focus on demonstrating the feasibility of using AI to predict TPMS fatigue life and securing funding for Phase II, which could expand the scope and impact of the research significantly.

The convergence of physics-based modeling, artificial intelligence, and advanced manufacturing represents a fundamental shift in how engineers approach materials discovery. Rather than relying on intuition, trial-and-error, or exhaustive testing, researchers can now use AI to navigate vast design spaces efficiently, identifying optimal solutions that might have remained hidden in traditional workflows. For industries ranging from aerospace to medicine, that capability could translate into lighter aircraft, more durable implants, and more resilient structures, all developed faster and at lower cost than ever before.