Causation-guided mechanism identification and interpretable reduced-order modeling of damage-driving grain-boundary stress in creep

TL;DR

This paper develops a causation-guided machine learning framework to identify key mechanisms governing grain-boundary stress in creep, enabling interpretable reduced-order modeling with robust generalizability.

Contribution

It introduces a causation entropy-based approach to distill physically meaningful descriptors for predicting grain-boundary stress in creep, enhancing interpretability and model performance.

Findings

Identified dominant characteristics: GB inclination, slip transmission, Schmid indicator, elastic mismatch.

The hierarchy and functions remain effective under multiaxial loading and in tricrystal systems.

Extracted functions improve surrogate model performance across machine-learning classes.

Abstract

Grain-boundary (GB) local stress is central to the initiation and evolution of long-term creep damage in polycrystalline superalloys. Owing to the high-dimensional nonlinear relationships between the GB stress response and multiple crystallographic, microstructural, and micromechanical characteristics, it remains challenging to identify the key characteristics governing GB stress and to elucidate their mechanisms of influence. Dislocation-climb-affected crystal-plasticity finite-element simulations of minimal grain clusters are combined with an integrated causation-guided machine-learning framework, in which mechanics-informed descriptors are analyzed by causation entropy to identify governing mechanisms and then distilled into a reduced-order regression form for interpretable prediction of GB normal stress. Among 18 physically motivated characteristics, the GB inclination angle, the slip transmission, the climb-related Schmid-type indicator, and the elastic-modulus mismatch are found to be dominant, revealing the coupled roles of interfacial geometry, crystallographic compatibility, creep stress relaxation, and micromechanical contrast. The identified characteristics hierarchy and functional representation remain effective under multiaxial loading and can be extended to tricrystal systems through physically interpretable nonlocal augmentation when a purely local description becomes insufficient, demonstrating strong physical consistency and robust generalizability across physical conditions. The extracted candidate functions also improve surrogate-model performance across multiple machine-learning model classes, providing supporting evidence for the physical relevance and efficiency of the identified representation. The proposed methods demonstrate strong potential for the development of interpretable machine-learning models and for the study of microscale nonlocal damage.