Speaker
Description
Accurately simulating localized plastic strain at geometric discontinuities, such as reentrant corners, remains a significant hurdle due to the computational demands of fully nonlinear modeling. While simplified elastic methods are faster, they fail to account for critical nonlinear effects. To bridge this gap, we introduce NeuberNet, a Multi-Task Nonlinear Manifold Decoder designed to map far-field displacement data from coarse elastic simulations to high-fidelity stress and strain distributions.
The framework operates on axisymmetric solid mechanics principles, assuming bilinear isotropic hardening and small-scale plasticity. By applying the substructuring principle, NeuberNet restricts nonlinear computations to localized regions near stress concentrators, drastically improving efficiency.
Our study establishes specific mesh density requirements for input simulations and validates the model’s capacity to detect when small-scale plasticity limits are exceeded. Furthermore, we demonstrate NeuberNet’s versatility by applying it to 3D scenarios involving axisymmetric geometries under asymmetric loading. The results indicate that NeuberNet offers a robust, high-speed alternative for analyzing localized plastic deformation in engineering components.