Multiscale Uncertainty Quantification of Woven Composite Structures by Dual-Correlation Sampling for Stochastic...

Area of Science:

• Materials Science
• Mechanical Engineering
• Computational Science

Background:

• Woven composites exhibit uncertainties from material properties to geometry, affecting structural performance.
• These uncertainties cause spatial correlations, leading to variable strength and damage patterns.

Purpose of the Study:

• To develop a multiscale uncertainty quantification framework for woven composites.
• To accurately propagate uncertainties from microscale to macroscale, considering spatial correlations.

Main Methods:

• Proposed a novel dual-correlation sampling approach based on multivariate random field (MRF) theory.
• Simultaneously captured spatial autocorrelation and cross-correlation for realistic material property representation.
• Validated the framework using in-plane tensile tests on woven composite structures.

Main Results:

• The dual-correlation sampling approach accurately predicted probabilistic mechanical responses and damage morphology.
• Demonstrated superior accuracy compared to traditional independent sampling methods.
• Highlighted the importance of capturing spatial correlations for reliable predictions.

Conclusions:

• The developed framework provides a robust method for uncertainty quantification in woven composites.
• Findings enhance structural reliability assessment and risk management in engineering.
• Emphasized the limitations of ignoring spatial correlations in composite material modeling.