Failure Analysis and Prediction of Hybrid Composite Laminate under Biaxial Tension using Artificial Neural Network
Hybrid composite laminates such as Boron/Epoxy or Glass/Epoxy show complex failure behaviour when subjected to biaxial tension on which it is difficult to accurately predict the failure load. In order to tackle this problem, we aim to develop an Artificial Neural Network model that can precisely for...
| 總結: | Hybrid composite laminates such as Boron/Epoxy or Glass/Epoxy show complex failure behaviour when subjected to biaxial tension on which it is difficult to accurately predict the failure load. In order to tackle this problem, we aim to develop an Artificial Neural Network model that can precisely forecast the failure load of hybrid composite laminates in biaxial tension. The angle orientation, volume fraction, and plate thickness (the three design factors) have been included as the key parameters. To systematically create the 17 unique combinations of the key parameters, a Box-Behnken Design framework was used, of which the combinations were then tested using finite element simulations in ANSYS. Three volume fraction configurations were considered: The angle orientation ranged from 0 to 45° and plate thickness varied at 0.5 mm, 1 mm and 3 mm intervals while it was 100% boron, 100% glass, or a hybrid of the two. A BBD framework was utilised to systematically generate 17 different combinations of the key parameters, which were subsequently leveraged in the FEA to generate a robust dataset to evaluate the failure behaviour under biaxial tension. The simulation results were used with ANN to make an accurate prediction of failure load. Results show optimal laminate design including angle orientation of 22.5°, hybrid material composition, and plate thickness of 1.75 mm. The highly accurate prediction was achieved by a developed ANN model trained with Levenberg-Marquardt algorithm with a cumulative R2 of 0.9949. The work presented here provides an advancement in understanding failure load behaviour in hybrid composite laminates in biaxial tension and also compares the accuracy of ANN predictions to finite element analysis results. © 2024 IEEE. |
|---|---|
| ISSN: | |
| DOI: | 10.1109/SCOReD64708.2024.10872670 |