Cross-linked elastomers, known for their exceptional flexibility, toughness, formability, and versatility, play a vital role in a wide array of engineering applications across aerospace, construction, transportation, marine, aeronautics, and automotive sectors. These materials are expected to maintain high performance throughout their service life, even when exposed to aggressive environmental conditions such as water infusion, temperature fluctuations, and ultraviolet (UV) radiation. These environmental factors pose significant challenges, as they can gradually deteriorate the material’s properties and reduce its overall durability.Among the most critical forms of environmental degradation are thermal aging under oxygen-deficient conditions and oxidative aging caused by elevated oxygen concentrations. The former typically results in uniform thermal degradation of the material, while the latter, known as diffusion-limited oxidation (DLO), induces spatially heterogeneous damage, primarily at the surfaces of the polymer where oxygen diffusion is more pronounced. Together, these two forms of aging represent fundamentally inverse degradation conditions—one being inert and volumetric, and the other being oxidative and surface-driven. Accurate prediction of the long-term behavior of elastomers under both these conditions is essential for designing reliable rubber components that resist early failure in service.To address this need, high-fidelity constitutive models are essential for simulating the effects of aging on the mechanical, thermal, and failure characteristics of cross-linked polymers. Historically, most aging models have employed hyperelastic constitutive laws coupled with single-kinetic degradation equations to model the evolution of material properties over time. While these approaches have been useful in capturing basic degradation trends, they often fall short in representing the complex interactions between microstructural evolution and macroscale mechanical behavior under realistic service conditions.This dissertation presents a comprehensive multi-physics modeling framework to capture the distinct and coupled degradation behaviors of cross-linked polymers under both diffusion-limited oxidation and inert thermal aging. These models incorporate the effects of oxygen diffusion, reaction kinetics, and thermally activated chain scission and cross-linking processes to simulate the evolution of polymer microstructure over time. By resolving the spatial and temporal development of aging parameters, the framework successfully reproduces both uniform and spatially heterogeneous degradation phenomena, providing critical insight into how these opposing environmental factors influence long-term material performance.The modeling approach is grounded in continuum mechanics and integrates finite strain theory with micro-mechanically motivated degradation mechanisms. Rubber elasticity is described across three scales: statistical mechanics at the microscale to account for molecular chain behavior, network-based phenomenological modeling at the mesoscale, and continuum theory at the macroscale. This multi-scale approach enables the representation of polymer network reconfiguration during aging, which is essential for predicting stiffness loss, permanent deformation, and eventual failure.While the core of this dissertation focuses on physics-based modeling of aging mechanisms, recent advancements in machine learning (ML), particularly physics-informed neural networks (PINNs), offer promising opportunities to enhance traditional modeling techniques. Although early generations of data-driven black-box models were limited by their need for large datasets and lack of physical constraints, hybrid approaches—where experimental macroscopic data are used to infer underlying microstructural behavior within physics-informed frameworks—have begun to bridge this gap. In this work, such techniques are explored in a supporting capacity to augment the predictive capabilities of the proposed aging models, without compromising the underlying physical consistency.By coupling the two degradation models and incorporating both mechanistic understanding and data-driven insights, this dissertation delivers a unified computational framework capable of simulating the long-term behavior of elastomeric materials under varied environmental exposures. The results not only enhance our understanding of how oxidative and inert aging conditions uniquely and jointly affect polymer durability, but also provide a practical foundation for the design of next-generation elastomers with improved resistance to environmental degradation.
Read
