Yuanxin Xiao

I am a Ph.D. student specializing in bio-inspired engineering, materials science, machine learning, and mechanical modeling. My research focuses on translating nature’s hierarchical structures—such as nacre and spicules—into advanced materials with exceptional toughness, energy absorption, and fracture resistance. By combining mechanics with data-driven approaches, I aim to advance next-generation materials and intelligent structures.

Key highlights of my work:

1. Integrated cohesive zone modeling, phase-field fracture mechanics, and high-fidelity finite element simulations to optimize layered systems for superior mechanical performance.

2. Leveraged physics informed deep neural networks to accelerate microstructural design, building surrogate models trained on large-scale FEA datasets for real-time predictive analysis.

3. Applied ML-driven frameworks in robotics, including adaptive locomotion systems where scalability and speed are critical.

My interdisciplinary approach bridges mechanics, data science, and biomimetics, driving innovations in materials design, robotics, and intelligent structures.

Executed advanced multiphysics modeling, AI-driven optimization, and crashworthiness simulations to enhance structural design and material performance. Pioneered Cohesive Zone Modeling (CZM) in HyperMesh for layered composites, achieving up to 56% improvement in energy absorption and fracture resistance. Implemented custom traction–separation laws to capture complex interfacial behaviors, reducing mesh sensitivity and improving solver convergence by 40%. Completed the full design-to-testing cycle by fabricating optimized multilayer structures via 3D printing, experimentally validating FEM predictions with 95% accuracy.

Developed a physics-informed neural network (PINN) to enhance finite element analysis of microstructural design in robotic components, training on over 1M simulation samples. Integrated PINN and FEA workflows into HPC clusters using Linux, Slurm, and Python, significantly reducing runtime and costs.

Advanced crashworthiness analysis of vehicle cabins and chassis using ANSYS LS-DYNA and Composite PrepPost, strengthening designs under frontal and lateral dynamic loads and improving energy absorption efficiency by 44%. Created an ML-based nonlinear rubber bumper material model, leveraging experimental data to enhance prediction accuracy of high-strain-rate impact behavior, improving safety outcomes and structural resilience.

Penn State University, Engineering Science and Mechanics