m (JSanchez moved page Draft Sanchez Pinedo 460606964 to Natarajan et al 2024a) |
|||
| (One intermediate revision by the same user not shown) | |||
| Line 3: | Line 3: | ||
For the manufacturing process simulation of fiber-reinforced polymer composites, f low simulations have to be performed at multiple spatial scales which govern the flow through the fiber structures. Repetitive multiscale flow simulations are computationally expensive and time-consuming. In order to speed up the multiscale simulation workflow, fast machine learning surrogate models or emulators could be used to replace one or more of the flow simulations. In this work, feature-based emulators and geometry-based emulators are developed using neural networks for predicting the permeability of 3D fibrous microstructures based on a reference dataset (doi:10.5281/zenodo.10047095). The best model achieved a mean relative error of 8.33% on the test set with a significantly faster inference time compared to a conventional simulator. | For the manufacturing process simulation of fiber-reinforced polymer composites, f low simulations have to be performed at multiple spatial scales which govern the flow through the fiber structures. Repetitive multiscale flow simulations are computationally expensive and time-consuming. In order to speed up the multiscale simulation workflow, fast machine learning surrogate models or emulators could be used to replace one or more of the flow simulations. In this work, feature-based emulators and geometry-based emulators are developed using neural networks for predicting the permeability of 3D fibrous microstructures based on a reference dataset (doi:10.5281/zenodo.10047095). The best model achieved a mean relative error of 8.33% on the test set with a significantly faster inference time compared to a conventional simulator. | ||
| + | |||
| + | == Full Paper == | ||
| + | <pdf>Media:Draft_Sanchez Pinedo_460606964pap_668.pdf</pdf> | ||
Latest revision as of 11:04, 23 October 2024
Abstract
For the manufacturing process simulation of fiber-reinforced polymer composites, f low simulations have to be performed at multiple spatial scales which govern the flow through the fiber structures. Repetitive multiscale flow simulations are computationally expensive and time-consuming. In order to speed up the multiscale simulation workflow, fast machine learning surrogate models or emulators could be used to replace one or more of the flow simulations. In this work, feature-based emulators and geometry-based emulators are developed using neural networks for predicting the permeability of 3D fibrous microstructures based on a reference dataset (doi:10.5281/zenodo.10047095). The best model achieved a mean relative error of 8.33% on the test set with a significantly faster inference time compared to a conventional simulator.
Full Paper
Document information
Published on 23/10/24
Submitted on 23/10/24
Volume Advances in machine learning for composite materials, 2024
DOI: 10.23967/eccomas.2024.023
Licence: CC BY-NC-SA license
Share this document
Keywords
claim authorship
Are you one of the authors of this document?