Uncertainty Quantification in Numerical Modelling of Polymeric Foam Composites Assisted by Machine Learning
The use of natural fibres to reinforce composites is becoming increasingly widespread, driven by the growing demand for renewable and sustainable material solutions. However, as composite structures become more complex, the need for accurate and reliable numerical modelling techniques becomes increa...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Proceeding |
| Language: | en |
| Published: |
2025
|
| Subjects: | |
| Online Access: | http://ir.unimas.my/id/eprint/49429/7/proceeding.pdf http://ir.unimas.my/id/eprint/49429/ https://www.ukm.my/cem/sdmms/flipbook.html |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The use of natural fibres to reinforce composites is becoming increasingly widespread, driven by the growing demand for renewable and sustainable material solutions. However, as composite structures become more complex, the need for accurate and reliable numerical modelling techniques becomes increasingly critical to predict their behaviour effectively. In this study, a combination of Finite Element Analysis (FEA) and Machine Learning (ML) is employed to numerically model and predict the material properties of a composite based on a polyester resin matrix reinforced with various fillers, including bagasse fibres and oil palm empty fruit bunch (OPEFB) fibres. The polymeric foam composite contains chaotically distributed fibres and randomly varying pore sizes and shapes. To address this uncertainty, an Interval Field (IF) approach is employed. A Representative Volume Element (RVE) of the composite structure is generated using the integrated FEA and IF approach to facilitate the homogenization procedure. The FEA results are compared and validated against previously obtained experimental data. Once satisfactory R-squared values are achieved, the validated data are used as training input for the Machine Learning model. The integration of Machine Learning into numerical modelling reduces the need for extensive experimental testing and helps minimise computational costs. This study highlights the potential of Machine Learning in the analysis and optimisation of polymeric foam composites. |
|---|