Scholarlink

<p>This study proposes a digital twin framework for rock triaxial mechanical testing. It integrates a micromechanical frictional-damage model with probabilistic assimilation using iterative local updating ensemble smoother (ILUES). The framework exploits the parameterization of the micromechanical model. ILUES mitigates parameter equifinality during inversion. The model captures microcrack growth and frictional slip. Seven physically interpretable parameters describe elastic, frictional, and damage responses. We assimilate stress–strain data across confining pressures using Bayesian updating. This enables online parameter calibration and comprehensive uncertainty quantification. Beishan granite case studies show rapid convergence, high predictive accuracy, and robustness. The approach outperforms the ensemble smoother with multiple data assimilation (ESMDA). The framework clarifies clearer microscale damage evolution and parameter interactions. It delivers reliable predictions from limited experimental data.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Uncertainty-Aware Micromechanical Digital Twin for Rock Behavior in Triaxial Compression Using the Iterative Local Updating Ensemble Smoother

  • 期刊论文

  • 发表日期：

    2025年12月8日

查看全文
  • Mingtao Hu,
  • Weiya Xu,
  • Ke Wang,
  • Changhao Lyu

摘要

This study proposes a digital twin framework for rock triaxial mechanical testing. It integrates a micromechanical frictional-damage model with probabilistic assimilation using iterative local updating ensemble smoother (ILUES). The framework exploits the parameterization of the micromechanical model. ILUES mitigates parameter equifinality during inversion. The model captures microcrack growth and frictional slip. Seven physically interpretable parameters describe elastic, frictional, and damage responses. We assimilate stress–strain data across confining pressures using Bayesian updating. This enables online parameter calibration and comprehensive uncertainty quantification. Beishan granite case studies show rapid convergence, high predictive accuracy, and robustness. The approach outperforms the ensemble smoother with multiple data assimilation (ESMDA). The framework clarifies clearer microscale damage evolution and parameter interactions. It delivers reliable predictions from limited experimental data.