Exploring emerging technologies for innovative structural solutions: deep energy method, fused filament fabrication, and hoof wall-inspired design

Chadha, Charul

Permalink

https://hdl.handle.net/2142/130110 Copy

Description

Title

Exploring emerging technologies for innovative structural solutions: deep energy method, fused filament fabrication, and hoof wall-inspired design

Author(s)

Chadha, Charul

Issue Date

2025-07-17

Director of Research (if dissertation) or Advisor (if thesis)

Jasiuk, Iwona

Doctoral Committee Chair(s)

Jasiuk, Iwona

Committee Member(s)

• Hutchens, Shelby
• Baur, Jeff
• Krishnan, Girish
• Koric, Seid

Department of Study

Mechanical Sci & Engineering

Discipline

Mechanical Engineering

Degree Granting Institution

University of Illinois Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

Hoof Wall-Inspired Design, Additive Manufacturing, Repair, Fused Filament Fabrication, Deep Energy Method

Abstract

Advancements in computational modeling, additive manufacturing (AM), and bio-inspired design are enabling the creation of structural materials with tailored mechanical properties and sustainable lifecycle performance. This dissertation investigates three synergistic thrusts that collectively advance the fields of Physics-Informed Neural Networks (PINNs), polymer repair via Fused Filament Fabrication (FFF) and material characterization of specimens printed using FFF, and multifunctional bio-inspired design. The first thrust focuses on enhancing the Deep Energy Method (DEM), a class of PINNs, as a mesh-free alternative to conventional finite element methods for solving boundary value problems. A two-loop optimization framework is developed to automate hyperparameter tuning using a Bayesian optimization algorithm. This framework eliminates the need for solution-dependent objective functions by adopting the principle of minimum potential energy as the loss function. Furthermore, Random Fourier Feature (RFF) mapping is introduced to reduce spectral bias during network training, thereby improving the accuracy and transferability of hyperparameters across various load cases, numerical resolutions, and geometries. The proposed approach achieves L2 displacement errors as low as 10⁻⁵ in two-dimensional linear elasticity problems and demonstrates generalizability to unseen geometries and loading conditions. The second thrust investigates the repairability of high-value thermoplastic components using FFF and the material properties of specimens printed using FFF. Experimental studies are conducted using FFF to evaluate the feasibility and adhesive strength of the repair process. The results indicate that print speed and print temperature are the dominant parameters that govern repair efficiency. Leveraging polymer healing theory, a simplified analytical model is developed to predict interlayer strength in amorphous polymers (e.g., ABS) with approximately 90% accuracy. Infrared thermography and micro-computed tomography (micro-CT) imaging are used to correlate thermal history, void morphology, and mechanical performance. Additionally, the effect of specimen size on mechanical properties is analyzed using compression testing and micro-CT for polycarbonate (PC) and thermoplastic polyurethane (TPU). It is shown that the elastic modulus in soft polymers like TPU significantly varies with specimen size due to void distribution, while stiffer polymers like PC exhibit size-invariant elastic properties. The third thrust focuses on the design of fracture- and impact-resistant materials inspired by the equine hoof wall, which combines high energy absorption and crack-deflecting features exhibited by tubular structures present in the equine hoof wall. The structure of the hoof wall was first studied using micro-CT. Results from micro-CT and prior literature helped develop analogies between 3D printable bio-inspired architectures and the hierarchical structure of the hoof wall. Systematic parametric studies assess the role of tubule geometry, orientation, and reinforcement on stiffness and energy absorption under various loading conditions. Finite element analysis and experimental validation reveal that the highest specific energy absorption is obtained when tubules are loaded axially, while loading perpendicular to the axis of the tubules assists in increasing energy absorption during crack deflection by arresting the crack and promoting directional crack deflection. The mechanism of crack deflection is studied in anisotropic printed materials to identify optimal tubular arrangement and geometry in specimens 3D printed using FFF. Overall, this dissertation advances the design of multifunctional materials and repair strategies by bridging machine learning-based simulation, polymer additive manufacturing, and biologically inspired structures. The tools and methodologies developed herein provide scalable frameworks for improving structural integrity, sustainability, and performance in both new and repaired components.

Graduation Semester

2025-08

Type of Resource

Thesis

Handle URL

https://hdl.handle.net/2142/130110

Copyright and License Information

Copyright 2025 Charul Chadha

Owning Collections