Ply-by-Ply Damage Prediction in Composite Structures
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional finite element analysis techniques struggle to accurately predict damage in composite structures due to their complex multi-layered nature, especially under ballistic loading conditions, as they typically rely on a single material failure model and fail to consider the diverse behavior of different materials and layers under various stresses and loads.
Innovation Solution
A computer-implemented method that combines multiple material failure models using deep learning to predict ply-by-ply damage in composite structures, aggregating predictions from separate models to provide a more accurate and robust damage assessment, incorporating geometric models and machine learning to quantify damage in each layer.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional finite element analysis techniques use a single material failure model, then the analysis is simple and computationally efficient, but the prediction accuracy of damage in composite structures is insufficient
Solution Approach 1:
The patent combines multiple material failure models (Hashin, Puck, Tsai-Wu, etc.) into a unified analysis framework that simultaneously evaluates different failure mechanisms. This merging of multiple models allows the system to capture diverse damage behaviors in composite structures while maintaining computational efficiency through integrated processing.
Solution Approach 2:
The patent applies composite material principles by creating a composite analysis approach that integrates multiple failure models, each suited for different material behaviors and failure modes. This composite modeling strategy enables accurate prediction of damage in multi-layered composite structures with varying ply orientations and material properties.
2Adaptability or versatility
If conventional FEA uses a single material failure model, then the computational process is fast, but it fails to account for diverse behavior of different materials and layers under various stresses
Solution Approach 1:
The patent segments the analysis by applying different material failure models to different plies and layers of the composite structure based on their specific material properties and expected failure modes. This segmentation allows each layer to be evaluated with the most appropriate failure criterion, improving adaptability while managing computational complexity through targeted analysis.
Solution Approach 2:
The patent implements local quality by assigning different failure models to different regions and layers of the composite structure based on their specific characteristics. Each ply can utilize the most suitable failure criterion for its material composition and orientation, enabling localized accurate prediction without requiring all models to run at full computational cost throughout the entire structure.
3Reliability
If multiple material failure models are applied to predict different types of damage, then the comprehensive damage assessment improves, but the complexity of aggregating predictions increases
Solution Approach 1:
The patent introduces an intermediary aggregation layer that systematically combines predictions from multiple material failure models. This intermediary component processes the diverse output from different failure models (Hashin, Puck, Tsai-Wu, etc.) and integrates them into a unified damage assessment, managing the complexity of aggregation through structured synthesis of multiple prediction sources.
Data Source
AI summary
A computer-implemented method that facilitates determining ply-by-ply damage in a composite structure comprises receiving, by a computing system, a geometric model that specifies geometric aspects of a composite structure. The geometric model facilitates the performance of finite element analysis (FEA). FEA logic of the computing system applies each of a plurality of material failure models (MFMs) to the geometric model to predict different types of damage to the composite structure due to an applied force or stress. Each MFM relates a force or stress applied to a material to a particular type of damage to the material. The machine learning logic predicts, based on each of the predicted different types of damage, an aggregate prediction of damage to the composite structure.