Related Experiment Video
Updated: Aug 25, 2025
09:11
High-resolution Volume Imaging of Neurons by the Use of Fluorescence eXclusion Method and Dedicated Microfluidic Devices
Published on: March 26, 2018
7.1K
A Statistical Porosity Characterization Approach of Carbon-Fiber-Reinforced Polymer Material Using Optical Microscopy
Sara Eliasson1,2,3, Mathilda Karlsson Hagnell4, Per Wennhage2,3
1Scania CV AB, SE-151 87 Södertälje, Sweden.
Materials (Basel, Switzerland)
|October 14, 2022
Summary
This study introduces an automated method using neural networks to analyze voids in Carbon-Fiber-Reinforced Composites (CFRP). This technique accurately characterizes material composition, aiding in understanding mechanical behavior and improving manufacturing processes.
Area of Science:
- Materials Science
- Mechanical Engineering
- Computational Science
Background:
- The increasing demand for lightweight materials in commercial vehicles drives the need for advanced composites like Carbon-Fiber-Reinforced Composites (CFRP).
- Manufacturing defects, particularly voids, significantly impact the mechanical properties and damage initiation in anisotropic CFRP materials.
- Accurate micromechanical characterization is crucial for understanding CFRP behavior.
Purpose of the Study:
- To develop and validate an automated optical microscopy approach for statistically characterizing porosity in CFRP laminates.
- To implement a neural network for efficient segmentation and labeling of material constituents (void, matrix, fiber) in micrographs.
- To determine constituent fractions and analyze their statistical significance in relation to material behavior.
Main Methods:
- Utilized optical microscopy for void characterization in CFRP laminates.
- Developed and applied a neural network for automated segmentation of micrographs, identifying voids, matrix, and fibers.
- Performed statistical analysis on extracted constituent fractions, including global and local void fractions.
Main Results:
- The neural network achieved high accuracy in constituent characterization with a minimal number of training images.
- Automated segmentation significantly reduced manual labor, demonstrating potential for design process efficiency.
- Significant differences were found between global and local void fractions, suggesting their importance in explaining material behavior variations.
Conclusions:
- The proposed automated optical microscopy and neural network approach offers an efficient and accurate method for CFRP void characterization.
- This technique can be valuable for repetitive tasks in design and manufacturing, saving time and resources.
- The identified significant differences in void fractions provide insights into the mechanical behavior of CFRP materials.
